METHOD FOR ESTIMATE THE DRIFT OF A FUEL PUMP OF A TURBINE ENGINE

Description

FIELD OF THE INVENTION

[0001] The invention relates to fuel circuits of turbomachines. STATE OF THE ART

[0002] During operation, a turbojet engine burns fuel in its combustion chambers. To this end, pumps, in particular a main pump, inject the fuel into the chambers via a hydromechanical metering unit (also known by the acronym FMU for the English expression fuel metering unit ) . The latter's role is to measure the correct amount of fuel to be injected into the engine and to recirculate the excess upstream of the fuel circuit. This hydromechanical block is linked to the injectors and the pump block and is controlled by an electronic box.

[0003] Here we are interested in the aging of the main fuel pump. Indeed, after prolonged operation, some of its characteristics change. This can be due to changes in clearances, the appearance of scratches or changes in tooth profiles linked to cavitation. In an extreme case, it would be possible that the pump could no longer provide enough flow when a flight restart is required, even if the flow rate would be sufficient for a ground start.

[0004] It is not possible to know the flow potential of the main pump unless a reproducible test procedure is provided and a flow sensor is integrated into the fuel circuit.

[0005] Currently, the pump's condition is monitored following the manufacturer's after-sales service. At regular intervals, the pump is tested to ensure proper operation. This test is carried out in the company's technical premises or by returning it to the supplier. Thus, to avoid reaching the pump's maximum potential, preventive maintenance is recommended, but it is not mandatory. If such maintenance is carried out, it requires testing facilities or the provision of spare parts, or even engine downtime. If deterioration is observed, the part is replaced without it being possible to determine the actual operating time before deterioration, or even the operating conditions that led to it.

[0006] On some turbojets, a check is carried out at start-up by observing the rotation speed required to set certain variable geometries in motion, but this technique returns a significant number of false alerts.

[0007] One aim of the invention is to more easily assess pump wear to avoid unnecessary removals, to constantly ensure operating margins and, if necessary, to raise a preventive alert concerning the pump.

[0008] Document US 2012 0219429 describes a method for estimating a drift of a turbomachine fuel pump. STATEMENT OF THE INVENTION

[0009] For this purpose, the invention provides a method for estimating a drift of a turbomachine fuel pump, the method comprising the following steps: measuring a pressure difference at terminals of a metering device supplied with fuel by the pump; calculating a position of a valve regulating the pressure difference; determining at least one estimated position of the valve from the pressure difference and at least one reference function giving the pressure difference as a function of the position; calculating at least one value representing a drift of the pump using the calculated position and the estimated position; and comparing the value with a predetermined threshold.

[0010] Thus, the invention takes into account the interaction between the flow rate passing through the pump and that recirculated through the regulating valve in order to determine the aging of the pump. It implements an analysis of the drift of the pump using the hydromechanical metering block and control means. It is based on the characterization of a law linking the pressure difference at the terminals of the metering device and the opening of the valve. During operation, a computer takes into account a value of the pressure difference measured at the terminals of the metering device to obtain the estimated opening of the valve and compare it with the calculated opening. This deviation reflects a difference between the flow rate obtained when the pump is first used and that obtained with the pump at the end of an operating period. By making it possible to estimate the extent of the aging of the pump, the invention provides savings in maintenance time.It also provides preventive warnings on the extent of aging in order to trigger removal if necessary without waiting for a breakdown to occur, such as an inability to restart the engine.

[0011] In one embodiment, the method comprises determining the one or more reference functions.

[0012] Advantageously, the determination of the function takes place after delivery of the turbomachine and before commercial operation of the turbomachine.

[0013] Indeed, the acquisition can be done by tests during reception of the turbomachine. These tests will take into account errors linked to variations in leakage flow from the rest of the system. If the precision is not sufficient, a correction coefficient can be applied.

[0014] Preferably: two estimated positions of the valve are determined from the pressure difference and two reference functions giving the pressure difference as a function of the position; two values ​​representing the pump drift are calculated using the calculated position and the estimated positions: and at least one of the values ​​is compared with the predetermined threshold.

[0015] We can predict that: at least one instantaneous value is calculated using the calculated position and the estimated position; and using the instantaneous value, at least one average value representing the pump drift is updated.

[0016] We can predict that: two instantaneous values ​​are calculated using the calculated position and the estimated positions; and using the instantaneous values, two average values ​​representing the pump drift are updated.

[0017] We can predict that the or each average is a moving average.

[0018] The invention also provides a turbomachine fuel circuit, the circuit comprising: at least one pump, a metering device configured to be supplied with fuel by the pump, the metering device comprising a sensor of a pressure difference at terminals of the metering device, a valve configured to regulate a pressure difference at the terminals of the metering device, and means capable of controlling the implementation of a method according to the invention.

[0019] In one embodiment, the pump is a positive displacement pump.

[0020] The invention also provides a turbomachine comprising a fuel circuit according to the invention. DESCRIPTION OF FIGURES

[0021] We will now present an embodiment of the invention by way of non-limiting example in support of the drawings in which: there figure 1is a diagram illustrating the fuel circuit within a turbojet engine for one embodiment of the invention; figure 2 is a diagram detailing this fuel system; the figure 3 is a diagram detailing the role of the valve in this circuit; the figure 4 shows curves of the functions giving the pressure difference as a function of the valve position; the Figure 5 is a flowchart of the steps of the method of the invention in the present embodiment; the figure 6 is a curve representing the %Q drift of the flow rate associated with the aging of the pump as a function of time; and the figure 7 is a view analogous to the figure 4 in another example. Engine Description

[0022] In this example, we consider an aircraft turbojet 1 forming a double-flow, double-spool turbomachine comprising a stator and a rotor. It has a main axis which serves as the axis of rotation of the rotor relative to the stator. It comprises from upstream to downstream, therefore from left to right on the figure 1 , a fan 2, a low-pressure compressor, a high-pressure compressor, one or more combustion chambers 8, a high-pressure turbine and a low-pressure turbine. These elements, with the exception of the fan, are part of a central part of the turbojet engine. Their moving parts rotating around the axis form the rotor.

[0023] The high-pressure compressor, the combustion chamber 8 and the high-pressure turbine form a high-pressure body which, together with the low-pressure compressor and the low-pressure turbine, defines a main airflow vein.

[0024] A fan casing 16 surrounds the fan 2 and the central portion so as to form a fan compartment and to define a secondary airflow vein.

[0025] In reference to the figure 1 , during its operation, the turbojet 1 burns fuel in its combustion chambers 8.

[0026] For this purpose, it includes a fuel circuit 7 comprising: at least one pump 20, a metering device 22 configured to be supplied with fuel by the pump, a valve 24 configured to regulate a pressure difference at the terminals of the metering device 22, and control means formed by an electronic box 26 and capable in particular of controlling the implementation of the method of the invention.

[0027] The pump 20 is here a volumetric pump. It forms a pumping unit capable of injecting the fuel into the combustion chambers 8 via the metering device 22. It is assumed here that the pumping unit comprises a single pump 20 but the invention applies in the same way by replacing the pump with a unit of several pumps.

[0028] The metering device 22 and the valve 24 form a hydromechanical metering unit 25 or FMU. The role of the FMU unit 25 is to meter the quantity of fuel to be injected into the engine 1 and to recirculate the excess further upstream in the fuel circuit, upstream of the pump. It is configured to receive fuel from the pump 20 and to communicate it to injectors 28.

[0029] The doser comprises a sensor 19 of a pressure difference at terminals of the doser.

[0030] In reference to the figure 2, the metering circuit 7 extends from a fuel tank 34 to the combustion chamber 28. It comprises a feed line for the metering device 11 called high-pressure, connecting an outlet of the pump 20 to the inlet of the metering device 22. The metering device is adapted to deliver a target mass flow rate to the injectors 28 of the combustion chamber 8 from an initial flow rate delivered to it by the pump via the line 11. The regulating valve 24 is able to regulate the flow rate delivered to the metering device 22. In particular, it is able to return to the inlet of the pump 20 an excess flow rate of fuel reaching the metering device 22, as a function of the pressure difference at the terminals of the metering device.

[0031] To do this, the valve 24 comprises a cylinder 31 and a slide 32 mounted to slide in the cylinder 31. We will describe one embodiment of the valve but other embodiments can be envisaged.

[0032] The spool 32 comprises at its ends two movement control sections 320, 321, which make it possible to control the position of the spool in the cylinder. A first control section 320 is connected to the high-pressure line 11. It therefore receives a fuel flow at the same pressure as the fuel delivered at the inlet of the metering device 22. A second control section 321 is connected to an outlet of the metering device.

[0033] In addition, the cylinder 31 comprises a first inlet port 310 connected to the high-pressure line 11, and a second outlet port 311 connected to a low-pressure return line 12, this line connecting the port 311 to the inlet of the pump 20. By port, we mean an evacuation opening putting an internal cavity of the valve 24 into communication with a fuel circulation line, regardless of the geometry of the opening. For example, an orifice may extend over all or part of the circumference of the cylinder.

[0034] The slide 32 is further configured to, depending on its position in the cylinder, selectively authorize or prohibit fluid communication between the orifices 311 and 310, and modulate the flow rate of fluid circulating from the line 11 to the line 12 via the orifices 310 and 311 of the valve 24, when this fluid communication is authorized.

[0035] In particular, an increase in the fuel pressure applied to the first control section 320 causes the cylinder to move towards the opposite section (on the figure 3 , downwards), which tends to free the orifice 11 to increase the flow recirculated towards the line 12.

[0036] In addition, the valve 24 comprises a return member 323, for example a spring bearing against the second control section 321, in order to return the spool to an equilibrium position when the pressure applied to the first control section decreases.

[0037] This allows the fuel pressure differential at the metering terminals to be mechanically regulated, by adapting the valve opening to the need to recirculate the excess flow delivered by the pump.

[0038] The pressure differential depends on the balance of the valve spool. The spring stiffness affects this balance. To ensure accuracy, it is sometimes useful to implement a system to compensate for the effect of the stiffness. This can be done by adding a chamber 33 whose pressure is modulated by openings 313 and 314.

[0039] Inside the cylinder 31 of the valve is defined a chamber 33, adjacent to an additional control section 322. The chamber is delimited on one side by this additional section, so that the fluid pressure contained by the chamber 33 can be exerted on the additional control section 322.

[0040] In addition, the cylinder 31 comprises one or more additional orifices 313, 314, these orifices opening into the chamber 33. The metering circuit further comprises a flow sampling line 13, one upstream end of which is connected to the high-pressure line 11 and one downstream end of which is connected to one of the orifices 313. The fuel metering circuit further comprises a flow return line 14, one upstream end of which is connected to the other orifice 314 and one downstream end of which is connected to the line 12.

[0041] This process makes it possible to modify, depending on the opening of the valve, the pressure of the chamber 33 and thus to correct the effect of the variation of the force induced by the spring on the regulation of the pressure differential at the terminals of the metering device. Therefore, contrary to the representation of the figure 4, it is possible to have a law of variation of the opening of the regulating valve as a function of the non-linear ΔP. This does not change the principle and application of the invention. Subsequently, we will assume as an example a linear law of variation of the ΔP as a function of the opening of the valve. Principle of the invention

[0042] We now present the principle of the invention.

[0043] The flow rate of the pump 20 is a function of its rotation speed to which is applied an efficiency dependent on the rotation speed, the pressure difference and the temperature.

[0044] This flow is divided, at a stabilized point, between an injected flow and a recirculated flow by the metering device 22. The recirculated flow passes mainly through the regulating valve 24 and depends on the pressure difference at the terminals of the pump 20. Thus, the position of the valve 24 reflects the flow delivered by the pump at given temperature, pressure and rotation speed conditions, as well as according to the injected flow.

[0045] Let us call X the quantity representing the opening section of the valve 24, expressed in units of surface. This quantity also represents the position of the slide 32 in the cylinder 31. We can estimate the actual opening X of the valve when a flow Q.(1+ε pump ) passes through it, as well as the calculated opening X calc when we think that the valve is crossed by a flow Q using the following formulas (1) and (2): X est = Q . 1 + ε pompes . ρ T − Wf K ∗ ρ T . Php − Plp X calc = Q . ρ T − Wf K ∗ ρ T . Php − Plp

[0046] Or : ρ is the density of the fuel in kg / l, W f is the mass flow rate injected by the metering device 22 in kg / h, P hp is the pressure at the outlet of the pump 20, P lp is the pressure at the inlet of the pump, and K is a coefficient specific to the geometry of the valve 24.

[0047] We do not have access to the flow rate value Q, but only to the rotational speed of the motor shaft. Therefore, we calculate Q using the reduction ratios and the displacement of the pump. However, if the pump wears out, its efficiency decreases. The functions we use to determine Q from the rotational speed are then no longer accurate.

[0048] When pump 20 drifts, its flow rate Q decreases. However: Q = W f + débit recirculé

[0049] Since the metered flow rate W f does not decrease, it is the recirculated flow rate that decreases, leading to a reduction in the opening X of the valve. In other words, when the pump is worn, the valve is more closed than when the pump was new.

[0050] The objective is then to determine as precisely as possible the actual position X of the valve to quantify the drift of the pump (ε pumps) resulting from its aging.

[0051] Let's take the example of a linear variation of ΔP as a function of the valve position. In reality, this is not necessarily the case, but the principle remains applicable whatever the form of this variation law. This is the central curve 29 on the figure 4showing the linear evolution of ΔP as a function of the valve opening X, and vice versa. We recommend that this curve 29 be acquired at the start of life for each engine (for example upon receipt of the equipment). This curve therefore represents the nominal operation of the system.

[0052] The balance of the valve is then written with the following formula (3): Δ P = F 0 + r é glage + k . X S

[0053] Or : ΔP is the control pressure difference, equal to the pressure difference across the metering valve, F0+setting is the force of the valve spring 323, k is the spring stiffness, and S is the pressure application surface.

[0054] ΔP depends essentially on X. If X varies, ΔP varies. Equation (3) allows calculating a position X of the valve as a function of the measured value ΔP in the case of a linear valve and corresponds to curve 29.

[0055] We will observe the pump drift through the prism of the variation it causes on the ΔP, the measurement of which is carried out on the FMU 25 block, to improve the dosage precision.

[0056] Let us assume an operating regime at the beginning of life which causes the valve to have a given opening X represented by point 30 on the curve. As the pump ages, the flow rate Q decreases, and therefore the actual opening X of the valve also decreases.

[0057] The measured value of ΔP decreases but without the system changing its calculated value X calc of the valve opening. We are then, for example, at point 35, located outside of curve 29. Therefore, the measured value of ΔP is not consistent with the one it should have for the calculated valve opening X.

[0058] Using reference curve 29, we then look for what actual opening X of the valve could explain the measured ΔP. This involves projecting the measured ΔP value to the left onto curve 30 to obtain the abscissa of point 36. This allows us to obtain a value X.

[0059] The equilibrium equation (3) is valid with the hypothesis of a valve without mechanical compensation, i.e. the hypothesis of a linear law between the pressure differential ΔP and the position X of the valve. Since the invention is not limited to this ideal case, we recommend acquiring curve 29 at the beginning of the engine's life to have the real law (including leaks and compensations) linking ΔP to X. Hence the need to project ΔP to the left on the reference curve 29 to obtain the real value of X.

[0060] With the estimated value X of the opening (or position) of the valve thus obtained, we can quantify the drift of the pump with respect to the reference curve 29. In fact, it is sufficient to apply the formula (4): %Q = Xest − Xcalc Xcalc + Wf K . Php − Plp

[0061] This formula follows from the combination of formulas (1) and (2).

[0062] In reality, it is difficult to work with simple intersections. This is because ΔP is not perfectly defined for a given valve opening X. For a given opening X, ΔP will be in a certain range around curve 29. This range is represented by the two reference curves 37 and 38 located on either side of reference curve 29 on the figure 4 . So, projecting to the left, we do not get a single value X is , but a range of values ​​[X is 1 ; X is 2] bounded by the two values ​​X is 1 and X is 2 obtained respectively by curves 37 and 38.

[0063] Similarly, we obtain a range of values ​​[%QB; %QH] of %Q delimited by the two values ​​%QB and %QH obtained with these two values ​​X is 1 and X is 2. The value %Q represents the flow drift associated with the aging of the pump. With each calculation of %Q, we obtain the range of values ​​[%QB; %QH] in which the real value of the pump drift is found.

[0064] This calculation then involves performing this calculation regularly to create a moving average [%QBm; %QHm] over the time span of the range [%QB; %QH]. The relevant information is not the calculated range of %Q values, but its evolution over time. It is through its evolution that we can observe a "drift" from its original behavior. The moving average [%QBm; %QHm] of the range is calculated over a constant sliding period T, which moves over time. Each time we obtain a new range of %Q values, we update the moving average.

[0065] This moving average of the %Q ranges is calculated, for example, as follows. Suppose that each %Q range includes an upper bound %QH and a lower bound %QB and that the average range forming the moving average includes an upper bound %QHm and a lower bound %QBm. We calculate %QHm as the average of the upper bounds %QH and %QBm as the average of the lower bounds %QB.

[0066] The average range is then compared to a reference range having predetermined upper and lower reference limits when the engine is put into service.

[0067] It is then sufficient to provide a predetermined threshold, for example 10%. If a value associated with the moving average of the %Q range, for example its lower limit %QBm, has a difference of more than 10% compared to the value of the lower reference limit, an alert is triggered indicating a threat to the quality of the pump's operation and requiring maintenance. It is based on the analysis of the desired precision that information on a maintenance request can be sent.

[0068] All of these elements can be acquired during startup sequences and / or continuously during flight.

[0069] A mention can be made here of variable geometry turbomachines. These are engines in which the compressor includes movable vanes which allow air to escape to the outside if necessary to avoid a pumping effect. These vanes are moved by cylinders themselves actuated by the fuel acting as a control fluid. The variable geometries are therefore also powered by the pump. When geometries are moved, the metered flow rate remains the same, the flow rate delivered by the pumps remains the same, but the flow rate recirculated by the valve decreases. From then on, we obtain peaks or troughs of the value X is .It is possible to estimate the flow rate consumed by the movement of the variable geometries and to add it to the recirculated flow rate, but it is preferable to acquire the measurement points implemented in the process on the ground or in cruise, that is to say when the positions of the geometries are known or at least stable and already included in the reference curves acquired at the start of life.

[0070] In general, gear pumps have a characteristic varying in the order of + / -5%, especially at low speed. During sizing, a design margin is taken by the supplier. This margin is of the order of 15% of the flow rate at low speed. Thus, an accuracy of the order of 10% on the pump flow rate estimated by the method of the invention would make it possible to detect a pump that has drifted to the minimum flow rate value ensuring proper operation.

[0071] By implementing sensors to improve metering accuracy and fuel density estimation, and using RPM acquisition, we estimated by studying the influencing variables that we obtain an accuracy of plus or minus 5% to detect this drift. This estimate can be refined by a system-level study, analyzing the different possible reference curve acquisition scenarios, as well as other factors that may influence the correct estimation of the pump flow rate law, such as the impact of transient regimes on the estimation. Providing a continuous analysis of the difference between the estimated pump flow rate and the reference flow rate allows the alert to be based on a moving average so as to cover the effects of these transients.

[0072] The control box 26 can also be used to improve the reception of the equipment, the accuracy of the dosage and its use on the engine. For this, sensors can be installed to correct the various uncertainty positions of the hydromechanical block (density, temperature, spring force, etc.).

[0073] As seen, by adding the pressure difference sensor 19 to the terminals of the metering device and using the speed, it is possible to estimate the flow rate of the pump 20 from the estimation of the valve opening. It is then possible to check the condition of the pump provided that the estimation accuracy is greater than the design margins of the pump. Implementation example

[0074] The implementation of the method of the invention for estimating a drift of the fuel pump 20 of the turbomachine 1 can therefore comprise the following steps executed in this order with reference to the figure 4 .

[0075] In step 40, the reference functions 37, 38 are determined after delivery of the turbomachine and before its commercial operation, the functions 37, 38 giving the pressure difference ΔP as a function of the position of the valve. For this purpose, sensors installed for other inputs on the FMU block (precision, failure detection, etc.) are used. In the non-linear case, the functions 37, 38 are each represented by a table of pairs of values ​​recorded in the control means.

[0076] The following steps are implemented during the operation of the turbomachine.

[0077] In step 42, the speed of pump 20 is determined, then its flow rate Q is calculated.

[0078] In step 44, using the sensor 19, a pressure difference ΔP is measured across the terminals of the metering device 22. Furthermore, the position X calc of the valve 24 is calculated using equation (2).

[0079] In step 46, two estimated positions X is 1, X is 2 of the valve are determined from the pressure difference ΔP and the two reference functions 37, 38, which amounts to projecting point 35 of the figure 4 on each of the curves 37, 38 to obtain the estimated positions.

[0080] In step 48, two instantaneous values ​​%QB and %QH are calculated using the calculated position X calc , the estimated positions X is 1, X is 2 and equation 4. %QB and %QH are the limits of the range of %Q.

[0081] In step 50, these values ​​%QB and %QH are used to calculate a moving average [%QBm; %QHm] of the range of %Q over a predetermined period of time T. For this, moving averages of %QB and %QH are determined. Moving averages of the limits of the range are thus calculated. During the first cycles, each average is calculated by taking into account all the corresponding available values. Then, when the initial period T has elapsed, each average is recalculated for the values ​​of the most recent period T to update it. Thus, using the instantaneous values ​​%QB and %QH, the two average values ​​%QB m are updated, %QH m representing the limits of the moving average of the range of the drift of pump 20.

[0082] Finally, at least one of the values ​​%QB m, %QH m is compared with the predetermined threshold. If the calculated deviation(s) exceed the threshold, an alert is triggered on the degree of aging of the pump. Supplements

[0083] Thanks to the acquisition of parameters and the characterization of the behavior of the valve 24 during reception of the equipment or the engine, the invention makes it possible to compare in real time the theoretical opening and the actual opening of the valve estimated through the measurement of ΔP and to deduce a deviation in the behavior of the pump.

[0084] There figure 6 gives an overview of the expected data and illustrates an estimate of what a %Q vs. time curve might look like. Each vertical band 64 is from a measurement and represents the range of values ​​[%QB; %QH] within which the actual value of the pump flow %Q drift lies. Over time, these ranges will shift downward.

[0085] The central transparent band 60 represents the relevant information. This is the moving average [%QBm; %QHm] of the pump drift ranges 60. It is representative of the measurements taken during the life of the motor. We then observe that the blue band gradually drifts compared to its initial value. We can thus determine an alert threshold.

[0086] In this example, curves 29, 37 and 38 come from a non-linear modeling of the valve which means that the bands do not all have the same amplitude whereas they would have had the same with a linear valve.

[0087] For the calculation of the moving average [%QBm; %QHm], %QBm is determined as the maximum of the %QB values ​​during the sliding period and %QHm as the minimum of the %QH values ​​during the same period.

[0088] Curve 62 represents the actual drift of the pump.

[0089] The time unit on the abscissa can be chosen in different ways. This allows the drift range to be determined and the average range to be updated once per flight or once every ten flights, for example.

[0090] To the figure 7 , we have illustrated an example of the shape of a non-linear curve of ΔP as a function of the position X. In this case, the same curve leads to obtaining several values ​​X1 to X5 for the estimated position of the valve. We therefore obtain several disjoint ranges for %Q or even a range in several disjoint sections. In this case, the step of calculating the average range [%QBm; %QHm] implements the determination of the intersection of the ranges during the period, which tends to obtain an average range [%QBm; %QHm] with a single section.

[0091] The invention does not require a specific position sensor on the valve 24 which would only be useful for this monitoring and would generate additional cost.

[0092] Many modifications can be made to the invention without departing from its scope.

[0093] In a simplified but less precise version, the process includes the following steps: calculation of a position X calc of the valve 24; determination of an estimated position X est of the valve from a measurement of the pressure difference and a reference function 29 giving the pressure difference ΔP as a function of the position; calculation of a value %Q m representing a drift of the pump 20 by means of the calculated position X calc and the estimated position X est; and comparison of the value with a predetermined threshold.

Claims

1. A method for estimating a drift of a turbine engine (1) fuel pump (20), the method comprising the following steps: - measuring a pressure difference (ΔP) at the terminals of a metering device (22) supplied with fuel by the pump (20); - calculating a position (Xcalc) of a valve (24) regulating the pressure difference; - determining at least one estimated position (Xest) of the valve based on the pressure difference (ΔP) and on at least one reference function (29, 37, 38) giving the pressure difference (ΔP) as a function of the position; - calculating at least one value (%Qm; %QBm, %QHm) representing a drift of the pump (20) by means of the calculated position (Xcalc) and of the estimated position (Xest); and - comparing the value with a predetermined threshold.

2. The method according to the preceding claims which comprises a determination of the reference function(s) (29, 37, 38).

3. The method according to the preceding claim, wherein the determination of the function occurs after a delivery of the turbine engine (1) and prior to commercial exploitation of the turbine engine.

4. The method according to one of the preceding claims wherein: - two estimated positions (Xest1, Xest2) of the valve are determined based on the pressure difference (ΔP) and on two reference functions (37, 38) giving the pressure difference (ΔP) as a function of the position; - two values (%QBm, %QHm) are calculated representing the drift of the pump (20) by means of the calculated position (Xcalc) and of the estimated positions (Xest1, Xest2); and - at least one of the values (%QBm, %QHm) is compared with the predetermined threshold.

5. The method according to one of the preceding claims, wherein: - at least one instantaneous value (%Q; %QB, %QH) is calculated by means of the calculated position (Xcalc) and the estimated position (Xest); and - at least one average value (%QBm, %QHm) representing the drift of the pump (20) is updated by means of the instantaneous value.

6. The method according to one of the preceding claims, wherein: - two instantaneous values (%QB, %QH) are calculated by means of the calculated position (Xcalc) and the estimated positions (Xest1, Xest2); and - two average values (%QBm, %QHm) representing the drift of the pump (20) are updated by means of the instantaneous values.

7. The method according to one of claims 5 or 6, wherein the or each average value (%Qm; %QBm, %QHm) is a moving average.

8. A turbine engine (1) fuel circuit (7), the circuit comprising: - at least one pump (20), - one metering device (22) configured to be supplied with fuel by the pump, the metering device comprising a sensor (19) of a pressure difference at the terminals of the metering device, - one valve (24) configured to regulate a pressure difference at the terminals, and - means (26) able to control the implementation of a method according to any one of the preceding claims.

9. The circuit according to the preceding claim wherein the pump (20) is a positive displacement pump.

10. A turbine engine (1) comprising a fuel circuit (7) according to one of claims 8 to 9.