WO2013110324A1

WO2013110324A1 - Dynamic compressor control with surge prevention

Info

Publication number: WO2013110324A1
Application number: PCT/EP2012/051062
Authority: WO
Inventors: Mehmet Mercangoez; Diego PARESCHI; Bjørnar BØHAGEN; Thomas BESSELMANN
Original assignee: ABB Technology AG
Current assignee: ABB Technology AG
Priority date: 2012-01-24
Filing date: 2012-01-24
Publication date: 2013-08-01
Anticipated expiration: 2014-07-24

Description

DYNAMIC COMPRESSOR CONTROL WITH SURGE PREVENTION

DESCRIPTION

The present invention relates to an apparatus for controlling the operation of a compression station.

In a further aspect, the present invention relates to a method for controlling the operation of a compression station.

Compression stations are widely used in many industrial applications, such as gas transportation pipelines, liquefied natural gas plants and the like.

A compression station typically comprises at least a rotational compressor, such as an axial or centrifugal compressor.

As is known, the physical/mechanical quantities that define the behaviour of a compressor follow fixed proportionalities, which may be represented using families of parametric curves. For example, the operating state of a compressor may be represented by compressor maps (Figure 3), in which parametric curves express the pressure ratio of the compressor (i.e. the ratio between the discharge pressure and the suction pressure) as a function of the gas mass flow, for different values of the compressor rotational speed.

Under normal circumstances, a compressor works in a permissible region of operation.

In the compressor map shown in Figure 3, such a permissible region is delimited by two boundary lines known as stonewall limit and surge limit.

The stonewall limit is generally defined as the maximum flow that a compressor can ingest at a certain speed.

Excessive flow across the compressor can cause vibration and fatigue failure that can damage the entire compressor over time.

On the other hand, generally, the stonewall limit can be efficiently maintained by adjustments on the downstream resistance to the flow or by high-flow alarms/trips in case of severe disturbances, such as the opening failure of a recycle valve.

The surge limit instead defines a critical boundary between a stable and an unstable operating region of the compressor.

As is known, surge is an instability phenomenon that arises in a compressor due to the interaction of this latter with the downstream load.

Surge generally occurs if the compressor mass flow decreases below a certain pressure- dependent limit (the surge limit), for a given rotational speed. The arising of surge phenomena reduces the life-time of the compressor and it may cause severe damages to the compression station.

Therefore, it is generally important that a compressor does not work in the instability region beyond the surge limit, during its operating life.

On the other hand, the efficiency of a compression station mainly depends on the operating point of the compressor, more precisely on the mass flow and the pressure ratio that is provided by the compressor.

Thus, a compression station achieves a high efficiency only if the compressor works in a region that is close to the surge limit.

Apparatuses/methods for controlling the operation of a compression station are therefore typically faced with the problem of maintaining the operating point of the compressor close to the surge limit without crossing it, so as to achieve high efficiency levels for the compression station without the occurrence of surge phenomena.

Most of the traditional apparatuses/methods for controlling the operation of a compression station are not capable of anticipating the occurrence of surge.

Therefore, they maintain the compressor in an operating area having a certain safety margin with respect of the surge limit.

The choice of such a safety margin is mostly based on empirical data and it may not be appropriate in certain operating conditions of the compression station (e.g. during transient states of the compressor).

Some conventional apparatuses/methods adopt derivative calculation models to assess the rate, at which the compressor is approaching the surge limit.

When it is determined that the operating point of the compressor may cross the surge limit, a feed-forward control action intervenes to ensure a safe operation of the compressor.

The magnitude of said feed-forward control action is mostly based on empirical data and experience of the control system engineers rather than the actual physic quantities governing the compression system.

Further, the practice has shown that it may be quite difficult to be properly choose the magnitude of the feed-forward action in some operating conditions of the compression station (e.g. during transient states of the compressor).

A further drawback of traditional control apparatuses/methods resides in that they generally intervene only on the regulation valves (in particular on the recycle valve) of the compression station in order to set the operating point of the compressor. Unfortunately, the regulation valves of the compression station are often affected by undesired phenomena, such as the so-called stick-slip effect, which cause the actual valve opening/closing not to be proportional to the control signals applied for operating them.

The actual operating state of said regulation valves may thus remarkably differ from the desired one at least for a certain period of time (typically few hundreds of ms).

A sole intervention on the regulation valves of the compression station may therefore not ensure a sufficiently fast and accurate response to potentially dangerous variations of the operating point of the compressor, due to such a non-ideal behaviour of the mentioned regulation valves.

Patent application WO2010/114786 describes a control apparatus, in which the variable-speed drive of the compressor is used for providing data indicative of mechanical quantities related to the operation of the compressor.

Such a control apparatus foresees to intervene only on the recycle valve in order to regulate the operation of the compressor.

The disclosed control apparatus is thus still subject to the remarkable drawbacks described above.

Patent application US2009/0253617 describes a control apparatus for a compression station, which intervenes only on the variable-speed drive of the compressor for regulating the operation of this latter.

In particular, the control apparatus reduces the torque or speed applied to the compressor when the operating point of this latter is moving towards an instability region, which is predefined in stored compressor maps expressing the compressor torque as a function of the rotational speed.

It has been observed that this solution may bring to remarkable inaccuracies of the control action particularly when the compressor does not work in a steady state.

The maps expressing the compressor torque as a function of the rotational speed may in fact not accurately represent the operating behaviour of the compressor during transients.

The main aim of the present invention is to provide an apparatus for controlling the operation of a compression station, and a method thereof, which enable the above-described drawbacks to be overcome.

Within the scope of this main aim, another object of the present invention is to provide a control apparatus and method, which allow to exert an effective, fast and accurate anti-surge control action on the compressor operation.

Another object of the present invention is to provide a control apparatus and method, which allow effectively mitigating the adverse impact of a possible non-ideal behaviour of the regulation valves of the compression station on the compressor operation.

Another object of the present invention is to provide a control apparatus and method, which are relatively easy to be implemented, at competitive costs with respect to currently available control apparatuses/methods.

The above main aim and objects, as well as other objects that will become apparent from the following description and attached drawings, are achieved according to the invention by an apparatus for controlling the operation of a compression station, according to the following claim 1 and the related dependent claims.

In a further aspect, the present invention also refers to a method for controlling the operation of a compression station, according to the following claim 8 and the related dependent claims. In a general definition the control apparatus and method, according to the invention, foresee to acquire input data indicative of the operating condition of the compression station, process said input data (preferably employing one or more calculation models), and generate output control signals for controlling both the operation of the compressor drive (preferably by means of a variable speed control signal and variable torque drive control signal) and the operation of one or more regulation valves of the compression station (preferably at least the recycle valve).

The control apparatus and method, according to the invention, are capable of intervening in a coordinated manner both on the compressor drive and on one or more regulating valves of the compression station in order to regulate the operation of the compressor.

The control apparatus and method, according to the invention, are capable of exploiting the different response time (to a control signal) of the compressor drive and of the regulation valves for exerting a control action that foresees subsequent time windows of intervention. As soon as a surge event is detected or predicted, the mentioned output control signals are generated and delivered to the compressor drive and the regulation valves.

The compressor drive, which is characterised by a shorter response time, is commanded to carry out a sequence of torque (or speed) variations, including temporary increase or decrease in torque (or speed), which are aimed at actively moving the compressor away from a surge condition.

Such a solution is quite effective as immediate intervention, since a variation of the torque (or speed) of the compressor determines immediately a small variation of the mass flow and only later a large variation in the pressure ratio.

Therefore, commanding the compressor drive to vary (positively or negatively) the torque (or speed) of the compressor allows to providing a fast response to undesired variations of the operating conditions of the compressor.

A control action can thus be already exerted during a first time window, in which the operation of the compressor cannot yet be influenced by an intervention of the regulation valves of the compression station.

The immediate intervention of the compressor drive can be advantageously tuned, for example by choosing appropriate torque (or speed) control profiles, to mitigate the adverse effects on the operation of the compressor due to the non-ideal internal dynamics (e.g. due to stick-slip phenomena) of the regulation valves.

Since variations on torque (or speed) of the compressor alone generally cannot provide a lasting protection against surge events, also one or more regulation valves of the compression station, in particular the recycle valve, are commanded to intervene for regulating the operation of the compressor.

Since said regulation valves have a longer response time, their intervention occurs during a second time window that follows the first time window, at which the compressor drive has already intervened.

The intervention of the regulation valves allows to permanently bringing back or maintaining the operating point of the compressor to/in a stable region of operation with a convenient safety margin with respect to the surge limit.

At this point, the compressor drive may be commanded to change the compressor torque (or speed) control profiles, according to the needs.

Further characteristics and advantages of the present invention will emerge more clearly from the description given below, referring to the attached figures, which are given as a non- limiting example, wherein:

Figure 1 schematically illustrates a compression station, the operation of which is controlled by the control apparatus and method, according to the present invention;

Figure 2 schematically illustrates an embodiment of the control apparatus, according to the present invention;

Figure 3 schematically illustrates an example of the compressor map;

Figure 4 schematically illustrates how the control apparatus and method, according to the present invention, intervene on the operation of the compression station.

With reference to the above-mentioned figures, a first aspect of the present invention concerns a control apparatus 100 for a compression station 50.

The compression station 50 comprises at least a rotating compressor 2 that may be, for example, a centrifugal compressor or an axial compressor.

Preferably, the compression station 50 comprises an input section, which includes an inflow pipe 8 that is operatively connected with a suction pipe 10 to provide the compressor 2 with a relatively low pressure input gas.

The compression station 50 preferably comprises also an output section, which includes an outflow pipe 16 that is operatively connected to a discharge pipe 12 to receive relatively high pressure gas from the compressor 2.

Further, the compression station 50 preferably comprises one or more recycle sections that comprises at least a recycle pipe 14 operatively connected between the discharge pipe 12 and the suction pipe 10.

Advantageously a cooler (not shown) can be located between the recycle pipe 14 and the suction pipe 10 to maintain a constant suction side pressure.

The compression station 50 comprises one or more regulating valves for regulating the mass flow of the gas flowing through it.

Preferably, a recycle valve 6 is operatively associated with the recycle pipe 14 for regulating the mass flow of the high pressure gas flowing back towards the suction pipe 10.

Alternatively, instead of a recycle valve 6 a blow off valve can be employed to direct the flow to the atmosphere instead of the suction pipe.

This last solution may be advantageously adopted if the medium being compressed is air itself or one of the main constituents of atmospheric gases such as nitrogen or oxygen.

Advantageously, an inflow valve 13 and an outflow valve 15 may be operatively associated respectively with the inflow pipe 8 and the outflow pipe 16.

The inflow valve 13 regulates the mass flow of the low pressure gas flowing towards the compressor 2 while the outflow valve 15 regulates the mass flow of the high pressure gas provided by the compressor 2.

The compression station 50 comprises a compressor drive 4, preferably of variable speed type.

The compressor drive 4 preferably comprises an electric motor (not shown) that is operatively associated with the compressor 2 to determine the compressor rotational movement.

As an alternative (not shown), the compressor drive 4 may comprise a driving arrangement operatively associated with a steam turbine.

Preferably, the input section of the compression station 50 comprises first sensor means 18, 20, 22 for providing first data PI, Tl, Fl indicative of physical parameters, in particular pressure, temperature and mass flow, of the gas flowing through the suction pipe 10. Preferably, the output section of the compression station 50 comprises second sensor means 24, 26 for providing second data P2, T2 indicative of the physical parameters, in particular pressure and temperature, of the gas flowing through the discharge pipe 12.

Alternative embodiments of the present invention may foresee a different positioning of the mentioned first and second sensor means 18, 20, 22, 24, 26.

For example, the mass flow sensor 22 may be conveniently positioned at the discharge pipe 12.

Nonetheless, it has to be noticed how the mentioned first and second sensor means are advantageously arranged to detect thermodynamic quantities related to the gas flowing through the compressor 2.

Preferably, the compressor drive 4 comprises third sensor means 41 to provide third data M indicative of physical quantities related to the operation of the compressor 2 (e.g. the torque applied to the compressor and/or the compressor speed) and/or the electric motor (e.g. the electric variables of the electric motor) that is possibly operatively associated to the compressor 2.

The control apparatus 100 may be a stand-alone device or integrated or operatively associated with other control systems of the plant, in which the compression station 50 is arranged.

Preferably, the control apparatus 100 is operatively associated with or embedded in a distributed control system (DCS) 200.

According to the invention, the control apparatus 100 comprises computerised means that are configured to acquire input data D indicative of the operating condition of the compression station 50, process said input data D and generate output control signals that comprise first control signals CI for controlling the operation of the compressor drive 4 and second control signals C2 for controlling the operation of at least one of the regulating valves of the compression station 50.

Preferably, the computerised means of the control apparatus 100 are configured to process the input data D according to one or more calculation models, advantageously of the multi- variable type.

Within the scope of the present invention, the term "computerised means" refers to one or more software programs, modules, routines, and/or instructions that are stored or up-loaded by the control apparatus 100 and are executed by a processing device (not shown) of the control apparatus 100, such as, for instance, a microcontroller or another digital processing device.

The term "calculation model" instead refers to a computing procedure for processing the input data D, according to one or more sets of characteristic equations.

As explained above (Figure 4), the mentioned first and second control signals CI, C2 are generated in such a way they impact on the operation of the compressor drive 4 and of said regulation valves respectively at a first and second time window Tl, T2, said first time window Tl preceding said second time window T2.

The effects of the intervention of the control apparatus 100 on the compressor drive 4 and the mentioned regulation valves thus occur in a predefined and coordinated sequence of time windows Tl, T2.

As a result of the intervention on the compressor drive 4, at the first time window Tl, the compressor 2 is advantageously moved away from the surge limit towards the permissible region of operation.

As a result of the intervention on the regulation valves, at the second time window T2 that follows the first time window Tl, the compressor 2 is stabilized to operate in the permissible region of operation.

Preferably, the mentioned input data D comprise data indicative of the thermodynamic parameters of the gas flowing through the compression station 50. These data are advantageously used by the computerised means 101, 102, 103 to calculate the torque (or speed) control profiles for operating the compressor 2.

Preferably, the input data D comprise the data PI, Tl, Fl, F2, T2 provided by the sensor means 18, 20, 22, 24, 26 and the data M provided by the sensor means 41.

Advantageously, the input data D may also comprise fourth data S indicative of desired set- points of operation for the compression station 50. Said set-points may comprise, for example, desired values for the mass flow and pressure ratio of the compressor 2.

The data S may be provided by the DCS 200 or by another device operatively associated with the control apparatus 100.

Preferably, sampling means (not shown) are operatively associated with or comprised in the control apparatus 100 to provide the input data D in a digital form, according to a predefined sampling period.

Preferably, the second control signals C2 are aimed at controlling the operation of at least the recycle valve 6 of the compression station 50.

In a preferred embodiment of the present invention, the control apparatus 100 comprises first computerised means 101, second computerised means 102 and third computerised means 103. The first computerised means 101 are configured to acquire the input data D and outputting the output control signals CI, C2. At each generic sampling instant k, the computerised means 101 advantageously acquire the input data D provided by the sensor means 18, 20, 22, 24, 26, 41 and, possibly, by the DCS 200, and store the acquired information in a memory or buffer (not shown) of the control apparatus 100.

At each generic sampling instant k, the computerised means 101 further advantageously acquire calculated control variables U from a memory or buffer (not shown) of the control apparatus 100 and provide the output control signals CI, C2 on the basis of the acquired output control variables U.

The mentioned output control signals are delivered to the compressor drive 4 (signals CI) and to the actuators (signals C2) that operate one or more regulating valves of the compression station 50.

As mentioned above, the control signals C2 are preferably delivered at least to the actuator that operates the recycle valve 6.

The second computerised means 102 are configured to calculate estimation data El indicative of the actual operating conditions of the compressor 2.

The computerised means 102 advantageously implement a state estimator that is aimed at providing, at the generic sampling instant k, an estimation of the operating state of the compression station 50.

In particular, the computerised means 102 provide an estimation of the operating state variables of the compression station 50 that are not comprised in the input data D, i.e. of the state variables that cannot be directly measured or acquired by the control apparatus 100. The first estimation data El are calculated by the computerised means 102 on the basis of a first non-linear multi-variable calculation model MODI, which advantageously describes the dynamic behaviour of the compression station 50 in discrete time.

The calculation model MODI may be expressed by the following equations:

x(k+l) = f(x(k), u(k)) (a) y(k) = g(x(k), u(k)) (b) where k is a generic sampling instant, x(k) is a state vector that collects the state variables representing the state of the compression station 50 at the instant k, u(k) is a control vector that collects the control variables for controlling the compressor drive 4 and one or more regulating valves (e.g. the recycle valve 6) at the instant k, y(k) is an input vector the collects the measured variables obtained from the input data D acquired at the instant k, f() is a non linear function describing the dynamic behaviour of the compression station and g() is a non linear function describing how the measured variables of the compression station 50 depend on the state variables of the compression station and on the control variables delivered by the control apparatus 100.

The estimation data El comprise an estimation vector (k), which collects the estimated state variables describing the state of the compression station 50 and which is provided by the computerised means 102, using the calculation model MODI, at each sampling instant k. The estimated state variables include at least the upstream and downstream pressures and the compressor flow.

Additionally, they may include the upstream and downstream temperatures, various valve opening states especially for the case of valves exhibiting non-ideal behaviour, and the compressor speed.

Preferably, the calculation model MODI is structured as an Extended Kalman Filter (EKF). This solution is advantageous since it allows to remarkably reduce the computational load, which is quite convenient in a real-time computing environment.

It is noticed that the mentioned functions f() and g() have to be known a priori, since they may be different for each possible configuration of the compression station 50.

To this aim, the control apparatus 100 preferably comprises fourth computerised means 105 for setting the calculation model MODI on the basis of data MAP derived from compressor maps, in which the pressure ratio of the compressor 2 is expressed as a function of the gas mass flow (Figure 3).

The computerised means 105 advantageously acquire the data MAP, define the calculation model MODI and store it in a memory of the control apparatus 100 before this latter starts operating.

The calculation model MODI is thus a predefined calculation model, which is used by the computerised means 102, 103 at each calculation cycle (i.e. at each sampling period).

It should be appreciated that acquisition of the data MAP from compressor maps, in which the pressure ratio is expressed as a function of the mass flow, allows to achieve a higher accuracy in the control action exerted by the control apparatus 100.

Since the data MAP are derived from this kind of compressor maps, they advantageously provide an accurate representation of the operating state of the compressor 2 both in a transient state and in a steady state.

This would not be possible if the data MAP were derived from other kinds of maps, e.g. maps expressing the compressor torque as a function of the rotational speed.

The data MAP may be provided by the DCS 200 (as shown in Figure 2) or another device operatively associated with the control apparatus 100. Alternatively, the data MAP can be directly extracted by the computerised means 105 on the basis of compressor maps stored in a memory of the control apparatus 100.

The third computerised means 103 are configured to calculate the output control variables U, on the basis of which the control signals CI, C2 are delivered by the control apparatus 100.

The control variables U are calculated by the computerised means 103 on the basis of a second linear multi-variable calculation model MOD2 that is obtained by executing a linearization of the first multi- variable calculation model, described above.

Initially, the computerised means 103 execute a linearization of the calculation model MODI

(equations (a), (b) above) around the estimated state of this latter, which is represented by the estimate state vector (k).

The computerised means 103 thus dynamically obtain the calculation model MOD2 from the calculation model MODI, at each sampling period.

The calculation model MOD2, which describes as well the dynamic behaviour of the compression station 50, may be expressed by the following equations:

(k+l) = Y(k) = A e(k) + B u(k) + fo (c)

(d) where A, B, C, D are suitable linearization matrices, f₀, go are suitable linearization vectors such that γ() and μ() are linear functions.

The calculation model MOD2 (equations (c), (d) above) is conveniently used to obtain predicted state variables (k+l), (k+p) for the compression station 50 over a finite time horizon p (i.e. up to a sampling instant k+p), depending of the current and predicted control variables u(k), ... , u(k+p).

The computerised means 103 are in fact configured to formulate an optimization problem for calculating the output control variables U, on the basis of the calculation model MOD2.

The adoption of a linear multi-variable calculation model (MOD2) allows the computerised means 103 to formulate and solve a quadratic optimization (QP) problem by means of QP solving schemes that are quite easy to be implemented in practice.

This allows to reduce the needed computational load with respect to using solving schemes dedicated to general non linear optimization problems.

The QP problem to be solved by the computerised means 103 may be expressed as, or similar to, the problem of:

finding the minimum values for the quadratic function (z-r)^T Q (z-r)

subject to the constraints A_in z <= h_in and A_eq z = B_eq £ (k) + h_eq. where:

z = [u(k), u(k+p), (k+l), £ (k+p)] is a vector collecting the predicted control variables u(k), ... , u(k+p) and the predicted state variables (k+l), ... , (k+p);

r = [u_r(k), u_r(k+p), £_r(k+l), t _^r(k+p)] is a vector collecting the reference values for the control variables and state variables of the compression station 50;

Q is a positive quadratic weight matrix;

A in and h in are a matrix and a vector corresponding to inequality constraints;

A eq, B eq and h eq are matrices and a vector corresponding to equality constraints. The values of the weight matrix Q may be conveniently selected for stating how important certain regulations (e.g. a pressure ratio regulation) are with respect to other available regulations (e.g. a mass flow rate regulation).

The values of the vector h in may be conveniently selected for expressing physical or desired limitations for the state variables or the control variables. For example, the constraints deriving from the need of not crossing the surge limit can be expressed by values of the vector h in.

Alternatively, to avoid infeasible solutions to the QP problem, desired limitations for the state variables can be expressed via so-called soft constraints, by introducing slack variables into the constraints to quantify violations of the very same, and by using quadratic weights in the cost function to penalize the slack variables.

The values of the vector h in may be conveniently selected for incorporating the equations of the second multi-variable calculation model (equations (c), (d) above) into the quadratic problem to be solved.

The values of the vector r can be advantageously calculated from the data S indicative of the desired set-points of operation for the compression station 50.

The incorporation of the predicted state variables (k+l), (k+p) into the vector z allows to conveniently reduce the computational load needed for formulating and solving the above described QP problem.

From the above, it is apparent that the illustrated QP problem is formulated as a problem of minimizing the distance between the predicted control variables u(k), u(k+p) and state variables (k+l), (k+p) from their corresponding reference values u_r(k), u_r(k+p), t_r(k+l), ... , £: _r(k+p).

The mentioned QP problem may however be formulated according to alternative approaches. For example, it may be formulated (in a so-called 5_u formulation) as a problem of minimizing the distance between subsequent predicted control variables u(k), ... , u(k+p). After having formulated the described QP problem, the computerised means 103 are configured to calculate a solution vector

u_opt(k+p), _opt(k+l), opt(k+p)] that solves the described QP problem.

The solution vector z_opt represents an optimal solution to the formulated QP problem, given the above illustrated constraints.

The computerised means 103 are configured to extract the output control variables U from the calculated solution vector z_opt, in particular from the set of variables u_opt(k), ... , u_opt(k+p). At the next sampling instant k+1, the computerised means 101, 102, 103 repeat the described cycle of acquiring and processing the input data D, calculating the output control variables U and providing the output control signals CI, C2.

In order to further reduce the computational load, the output control variables U, which have been calculated for controlling at least a regulation valve of the compression station 50, may be kept constant for one or more sampling periods while the output control variables U, which have been calculated for controlling the compressor drive 4, are conveniently calculated at each sampling period.

Additional solutions, such as the adoption of soft constraints, may be conveniently used to reduce the computational load.

In a further aspect, the present invention relates to a method for controlling the operation of the compression station 50.

The method, according to the invention, comprises the steps of:

acquiring the input data D;

processing the input data D;

providing output control signals comprising first control signals CI for controlling the operation of the compressor drive 4 and second control signals C2 for controlling the operation of at least one of the regulating valves of the compression station 50, preferably for controlling at least the recycle valve 6.

Advantageously, the first and second control signals CI, C2 are provided in such a way they impact on the operation of the compressor drive 4 and of said regulation valves respectively at a first and second time window Tl, T2, said first time window Tl preceding said second time window T2.

In a preferred embodiment of the present invention, the input data D are processed according to one or more multi-variable calculation models.

Preferably, the step of processing the input data D comprises the steps of:

acquiring the input data D; calculating estimation data El indicative of the actual operating conditions of said compressor, said estimation data El being calculated on the basis of a first multi- variable calculation model MODI ;

calculating output control variables U, on the basis of a second linear multi-variable calculation model MOD2 that is obtained by executing a linearization of said first multi- variable calculation model on the basis of said estimation data;

providing the output control signals CI, C2 on the basis of said output control variables U.

Preferably, the step of calculating the output control variables U comprises the steps of:

executing a linearization of the calculation model MODI in order to obtain the calculation model MOD2;

formulating a quadratic optimization problem for calculating the output control variables U;

calculating the solution vector z_opt for solving said quadratic optimization problem; extracting the control variables U from the solution vector z_opt.

Preferably, the steps of the method, according to the invention, are repeated at each sampling period of the input data D.

Preferably, the method, according to the invention, comprises the preliminary step of setting the calculation model MODI on the basis of the data MAP related to compressor maps, in which the pressure ratio of the compressor 2 is expressed as a function of the mass flow of the gas flowing though it.

It has been shown in practice that the control apparatus and method, according to the invention, achieves the aim and objects stated above.

The control apparatus and method, according to the invention, intervene on both the compressor drive 4 and on one or more regulation valves of the compression station 50 in a coordinated manner.

The adoption of effective control techniques to generate the control signals CI, C2 and the delivering of the generated output control signals to different kind of devices regulating the operation of the compressor 2 (e.g. to both the compressor drive 4 and the recycle valve 6) allows to properly exploit the properties of these different kind of regulators that are present in the compression station, while mitigating the adverse impact of their possible non-ideal behaviour on the operation of the compressor.

Despite the fact that multi-variable control techniques are generally considered as unsuitable for real time control applications, such as anti-surge control applications, the control apparatus and method, according to the invention, allow to exert an effective, fast and accurate anti-surge control action on the compressor operation.

Thanks to the adoption of concurrent calculation models, the control apparatus and method, according to the invention, require an acceptable computational load, suitable for a real time control application, and they are relatively easy to be implemented in practice.

The control apparatus and method, according to the invention, can thus be practically implemented to compression stations of different types (even quite structurally different from that one shown in Figure 1), at competitive costs with respect to the solutions that are already present in the state of the art.

Claims

An apparatus (100) for controlling the operation of a compression station (50), said compression station comprising:

at least a rotating compressor (2);

one or more regulating valves (6, 13, 15) of the gas flowing through said compression station;

a compressor drive (4) operatively associated with said compressor;

characterised in that it comprises computerised means (101, 102, 103) that are configured to:

acquire input data (D) indicative of the operating condition of said compression station;

process said input data (D);

provide output control signals comprising first control signals (CI) for controlling the operation of said compressor drive (4) and second control signals (C2) for controlling the operation of at least one of said regulating valves (6, 13, 15).

An apparatus, according to claim 1, characterised in that said first and second control signals (CI, C2) impact on the operation of said compressor drive and said regulation valves respectively at a first and second time window (Tl, T2), said first time window (Tl) preceding said second time window (T2).

An apparatus, according to one or more of the previous claims characterised in that said computerised means (101, 102, 103) are configured to process said input data (D) according to one or more multi- variable calculation models (MODI, MOD2).

An apparatus, according to claim 3, characterised in that it comprises:

first computerised means (101) for acquiring said input data (D) and providing said output control signals (CI, C2);

second computerised means (102) for calculating estimation data (El) indicative of the actual operating conditions of said compressor, said estimation data (El) being calculated on the basis of a first multi-variable calculation model (MODI); third computerised means (103) for calculating output control variables (U), said output control variables being calculated on the basis of a second linear multi- variable calculation model (MOD2) that is obtained by executing a linearization of said first multi-variable calculation model (MODI) on the basis of said estimation data, said first computerised means (101) providing said output control signals (CI, C2) on the basis of said output control variables (U).

5. An apparatus, according to claim 4, characterised in that said third computerised means (103) are configured to:

execute a linearization of said first multivariable model (MODI) in order to obtain said second multi-variable calculation model (MOD2);

formulate a quadratic optimization problem for calculating said output control variables (U);

calculate a solution vector (z_opt) that solves said quadratic optimization problem; extract said output control variables (U) from said solution vector (z_opt).

6. An apparatus, according to one or more of the claims from 4 to 5, characterised in that it comprises fourth computerised means (105) for setting said first multi-variable calculation model (MODI) on the basis of data (MAP) related to compressor maps, in which the pressure ratio of said compressor (2) is expressed as a function of the mass flow of the gas flowing through said compressor.

7. An apparatus, according to one or more of the previous claims characterised in that said computerised means (101, 102, 103) generate said second control signals (C2) for controlling the operation of at least a recycle valve (6) of said compression station (50).

8. A method for controlling the operation of a compression station, said compression station comprising:

at least a rotating compressor (2);

a compressor drive (4) operatively associated with said compressor;

characterised in that said it comprises the steps of:

acquiring input data (D) indicative of the operating condition of said compression station;

processing said input data (D);

providing output control signals comprising first control signals (CI) for controlling the operation of said compressor drive (4) and second control signals (C2) for controlling the operation of at least one of said regulating valves (6, 13, 15).

9. A method, according to claim 8, characterised in that said first and second control signals (CI, C2) impact on the operation of said compressor drive and said regulation valves respectively at a first and second time window (Tl, T2), said first time window (Tl) preceding said second time window (T2).

A method, according to one or more of the claims from 8 to 9, characterised in that said input data (D) are processed according to one or more multi-variable calculation models (MODI, MOD2).

A method, according to claim 10, characterised in that the step of processing said input data comprises the steps of:

acquiring said input data (D);

calculating estimation data (El) indicative of the actual operating conditions of said compressor, said estimation data (El) being calculated on the basis of a first multi-variable calculation model (MODI);

calculating output control variables (U), said output control variables being calculated on the basis of a second linear multi-variable calculation model (MOD2) that is obtained by executing a linearization of said first multi- variable calculation model (MODI) on the basis of said estimation data;

providing said output control signals (CI, C2) on the basis of said output control variables (U).

A method, according to claim 11, characterised in that the step of calculating said output control variables (U) comprises the steps of:

executing a linearization of said first multivariable model (MODI) in order to obtain said second multi-variable calculation model (MOD2);

formulating a quadratic optimization problem for calculating said output control variables (U);

calculating a solution vector (z_opt) that solves said quadratic optimization problem;

extracting said output control variables (U) from said solution vector (z_opt). A method, according to one o more of the claims from 11 to 12, characterised in that it comprises the step of setting said first multi-variable calculation model (MODI) on the basis of data (MAP) related to compressor maps, in which the pressure ratio of said compressor (2) is expressed as a function of the mass flow of the gas flowing though said compressor.

A method, according to one or more of the claims from 8 to 13, characterised in that said second control signals (C2) are generated for controlling the operation of at least a recycle valve (6) of said compression station (50).