Patent Application
20160377333
This application claims priority to European Patent Application No. 15173519.8, filed on Jun. 24, 2015. The entire disclosure of the above application is incorporated herein by reference.
This invention relates in general to performance predicting, control, fault detection and diagnostic of refrigeration systems, and in particular to the use and the calibration of system performance models for such purpose.
Since the maintenance and the operation of refrigeration systems is quite expensive, there exists the aim of continuously monitoring and improving the reliability and the efficiency of refrigeration systems. Mechanical refrigeration systems that contain at least one compressor consume high amounts of energy during operation, and are subject to a number of highly volatile parameters which influence the reliability, the efficiency and capacity of the refrigeration system. Therefore, it is important to continuously monitor and control the performance of refrigeration systems.
In the prior art, a great variety of different ways are known how the operation of refrigeration circuits containing at least one compressor can be monitored and controlled starting from simply comparing sensed system values among them or to predetermined thresholds values, up to more sophisticated model-based methods using transfer functions to express the system's input/output relationship.
Fault detection and diagnosis of refrigeration systems is commonly handled by monitoring system's components operating parameters and following their evolution in time, where a component could be any component that is comprised within the system such as, for example, a compressor, or an expansion valve, etc. In addition to traditional fault detection methods as disclosed in EP 0 217 558 B1, methods based on performance prediction of refrigeration system components can be used. For example, U.S. Pat. No. 6,799,951 B2 describes a compressor fault or failure detection method based on the use of a predetermined compressor dataset such as performance rating curves to predict the value of a compressor operating parameter and compare it to an actual sensed value. Also, U.S. Pat. No. 6,981,384 B2 illustrates the use of a pre-programmed controller dedicated to detect non-adapted charge, i.e. system under-charge or over-charge, of refrigerant in a given system condition based on the comparison between the measured liquid sub-cooling and a pre-determined sub-cooling value defined as function of the operating mode and the characteristics of the system. Also, US 2005/0126190 A1 discloses a similar method based on suction superheat monitoring. EP 0 883 047 B1 shows a method dedicated to Electrical Expansion Valve, EEV, operation monitoring based on neural network theory. The aim of this method is to generate an algorithm based on the information generated by a plurality of sensors located in the system and to produce a computed value of the EEV position. As compared to other methods that are known in the prior art, this method does not rely on pre-determined parameters but has a very limited scope. The EEV control method described in US 2013/0205815 A1 is based on transfer function theory and does neither rely on pre-determined parameters. However, application of such a method is computationally demanding and of limited scope as well.
Among others, monitoring and controlling the performance of refrigeration systems also includes to measure the efficiency of the system and to compute performance indexes like Coefficient of Performance, COP, or Energy Efficiency Ratio, EER, for fault detection purposes, e.g. to identify potential efficiency degradation of the system or to check if the system is performing within its manufacturer's specifications, or performance monitoring purposes, e.g. to estimate operating costs and allow the owner to make decision about use and operation of its installation. However, measuring the efficiency of refrigeration systems is generally challenging, e.g. because there are multiple parameters to monitor, costly, e.g. because there are numerous sensors and loggers needed and, somehow, poorly accurate, especially for air systems. For example, U.S. Pat. No. 6,701,725 B2 describes the use of a widely known compressor performance model to estimate the capacity and the power of an actual refrigeration system based on published generalized, or default/standard compressors performance data, e.g. ARI Standard 540 or compressor manufacturer's tables, in order to draw conclusions about the performance of the actual refrigeration system, e.g. capacity, power, COP, EER, seasonal performance, etc. U.S. Pat. No. 8,775,123 B2 shows another simple method to estimate the coefficient of performance of a refrigeration system based only on enthalpy calculations and limited sensing of system parameters.
For controlling the operation of refrigeration circuits, the Electronic Expansion Valve, EXV, control is commonly handled based on standard closed-loop control algorithms, e.g. PI or PID control, intending to maintain a sufficient amount of superheat at compressor(s) suction in order to avoid excessive amount of liquid entering the compressor(s) that could cause failure of the compressor(s). Many methods exist to improve the robustness and the accuracy of superheat control, such as auto-tuning or adaptive-tuning methods as explained in U.S. Pat. No. 5,506,768 A, whereas WO2008/147828 A discloses another type of control based on the use of fuzzy logic method. However, these control methods generally do not allow operating with a compressor suction superheat of less than 5K to 7K, leading to suboptimal operation of the evaporator, and so, of the refrigeration system. It is well known that, controlling more accurately the superheat allowing operation at low, i.e. positive, or zero superheat, i.e. typically, 3K to 0K, would lead to higher system efficiency. Moreover, as explained in U.S. Pat. No. 4,878,355 A it is also well known that operating with limited amount of liquid droplets at the suction of the compressor may be beneficial to cool down the compressor and to extend its nominal operating range/envelope. However, such an operation, close to system reliability limits, requires high levels of accuracy and robustness that traditional control methods cannot provide. Therefore, EP 0 237 822 B1 describes a method to control the expansion valve opening based on the measurement of compressor discharge superheat and its comparison to predetermined degree of superheat estimated based on a relationship between compressor suction and discharge superheat values. WO 2009/048466 A1 shows a similar approach also based on the use of a relationship between compressor suction superheat and discharge temperature. U.S. Pat. No. 6,318,101 B1 describes a control method targeting evaporator pinch minimization and monitoring the discharge temperature deviation versus a predetermined setpoint that is stored in the controller to protect the compressor from liquid slugging. U.S. Pat. No. 7,509,817 B1 shows a linear expansion valve control method based on suction and discharge superheat measurement and characterized by the successive use of two different control methods making use of predetermined parameters representative of compressor(s) capacity(ies) and applied after a predetermined period of time. U.S. Pat. No. 6,711,911 B1 discloses another expansion valve control method based on the comparison of the actual (sensed) compressor discharge temperature and a theoretical value of the compressor discharge temperature that would correspond to a desired low-superheat operation. The calculation of the theoretical discharge temperature is based on some pre-determined coefficients/parameters characterizing the compressors in use in the refrigeration system. U.S. Pat. No. 8,096,141 B2 shows another method of controlling superheat operation of refrigeration systems relying on estimating the actual suction flow rate (based on some known or pre-determined characteristics of the valve) and adapting to the expansion valve opening to match a calculated desired suction flow rate set point, corresponding to a desired suction condition set point. U.S. Pat. No. 7,290,402 B1 shows an expansion valve control method based on an experimentally predetermined relationship between suction superheat error and valve opening. Pre-determined lookup tables are also used in US 2013/0174591 A1 to establish a superheat set point based on current operating conditions of the system.
As shown above, most of the monitoring, control and diagnostic methods described in the literature generally make use of nominal characteristics, predetermined relationships and/or reference standardized rating performance. Such data is usually generated based on the performance of one given system or component, at one given time. It is obvious that such pre-determined data cannot be considered as an accurate representation of the behavior of any, even similar, system or component since it does not take into account the effects caused by system's components manufacturing tolerances, break-in, ageing and application or system tolerances. However, for accurate monitoring, control, fault detection and diagnostic of the refrigeration system, it is essential to better adjust the applied performance prediction model during operation to account for components performance variability and variations.
Manufacturing variability of refrigeration system components may significantly impact components performance. Manufacturing tolerances vary from one component to another and from one manufacturer to another. Values of these tolerances and variability ranges are most of the time not available in the public domain and are part of the exclusive know-how of the component manufacturer.
A break-in effect is generally observed during the first hours of operation of a system component, wherein the break-in can be defined as the period of time until the system's components have reached stable performance level. Depending on the various parameters, such as the type of component, the technology, the size, the operating conditions, etc., the break-in may last for a couple of hours up to a few days. Break-in parameters, such as duration, are generally not available in the public domain and require a significant amount of test data to be identified. Then, these characteristics are also generally not available or are part of the exclusive know-how of the component manufacturer.
Components ageing consist in the evolution generally degradation of components operating performance with time. Such ageing and the resulting performance variation range may depend of many factors such as the time of use, the operating conditions, and the type of component used.
Usually, significant impact of ageing may occur after a relatively long running period. Indeed, most of system components are designed to meet life expectancy requirements of several years. Once again, accurate characterization of components ageing generally requires significant amount of data and very detailed knowledge of components behavior.
Methods where the effect of manufacturing tolerances, break-in and ageing of main component can be precisely characterized, anticipated and corrected offer new opportunities for more accurate performance and operation prediction, monitoring, control and fault detection. Indeed, as an example, the prior-knowledge of these tolerances allows more realistic definition of control and detection bands as well as more robust adjustment of control and detection algorithms.
Application or system tolerances may also have a certain impact on system performance and operation. Application tolerances include all the aspects that may vary between one system using a defined set of components and another system using the exact same set of components. This may include refrigeration circuit arrangement and related pressure drops, as well as quantity of oil and refrigerant in the system, presence of insulation on the components, presence of sound insulation cover on the compressor, etc. However, some of these influences are generally of smaller influence and may be easily compensated/corrected or taken into account if sufficient amount of sensors is used, e.g. pressure drops between two components.
Also, it is important to notice that current methods are often sub-optimal since they do not intend to characterize the system as a whole but, generally, consider system components individually, such as compressors, valves, etc., rarely leveraging on the close dynamic coupling existing among the components of a given refrigeration system.
Therefore, there is a need for efficient techniques to enhance the accuracy and the robustness of prediction models to be used for performance predicting, control, diagnostic and fault detection of refrigeration systems.
This need is fulfilled by the subject-matter of the independent claims.
According to the invention, a method of performance model cross-mapping in a refrigeration circuit containing at least one compressor and an expansion valve comprises the steps of: measuring one or more circuit parameter values of the refrigeration circuit, calculating a discharge line temperature, Tpm, with a first performance model as a function of at least one of the one or more measured circuit parameter values and comparing the calculated discharge line temperature, Tpm, to a measured discharge line temperature, Tmeas, from the refrigeration circuit to obtain a first differential value, ΔT, calculating a first flow, Mpm, with the first performance model as a function of at least one of the at least one or more measured circuit parameter values, calculating a second flow, Mevm, through the expansion valve with a second performance model for the expansion valve as a function of at least one of the at least one or more measured circuit parameter values, comparing the first flow, Mpm, to the second flow, Mevm, to obtain a second differential value, ΔM, and evaluating the first differential value, ΔT, and the second differential value, ΔM.
The refrigeration system which contains at least one compressor and at least one expansion valve consists in a closed-loop refrigeration circuit in which a refrigerant fluid flows and where the at least one compressor is dedicated to compress the refrigerant. As described above, the refrigeration circuit contains at least one compressor. However, depending on the application, the refrigeration circuit may also contain multiple identical or different, fixed capacity or variable capacity compressors.
Such a circuit may also contain a condenser which receives hot refrigerant gas from the compressor, wherein the gas is condensed in the condenser, and then fed through at least one expansion valve to the evaporator. By means of the expansion valve, the refrigerant flow into the evaporator can be controlled. The expansion valve used in the refrigeration circuit can be an electronic expansion valve, EXV, where a step motor is used to control the valve opening, and thereby controls the flow of refrigerant entering the evaporator. For example, the step motor can control the flow in response to signals received by an electronic controller.
The first performance model of the refrigeration circuit can be any model capable of modeling the behavior of the refrigeration circuit such as a black-box regression-type model, a white-box deterministic model that relies on physical relationships or a grey-box model that uses both approaches. Also, the second performance model can be any model capable of modeling the behavior of the refrigeration circuit such as described above with regards to the first performance model.
For example, a black-box regression-type model suitable for modeling the performance of the refrigeration system is defined in EN 12900/AHRI540 which can be used to directly compute compressor performance, such as typical power consumption, suction flow, current and capacity, based on compressor operating values such as the various pressure and temperature values that are sensed by means of sensors arranged in various locations of the circuit. Grey-box models that would be suitable for modeling the performance of the refrigeration system rely on the definition of standard efficiency indexes, such as compressor isentropic efficiency, electronic expansion valve characteristic factor, and the use of simple regressions to estimate the value of such indexes as function of the operating condition of the system, e.g. pressures and temperatures and components operating status, e.g. compressor speed, valve motor position. Parameters of these models may generally be identified for a given type of system/component and can be re-used to estimate the performance of similar equipment.
More sophisticated regression models are also suitable for modeling the performance of the refrigeration system and its components. These kinds of models employ a generalized notion of transfer functions to express the relationship between the input, the output and the noise in the circuit based on measured values from the circuit. Practice shows that AutoRegressive models with or without external input, e.g. like ARMA, ARX, ARMAX are specifically suitable for modeling the behavior of refrigeration systems containing at least one compressor, because these kinds of models can be easily adjusted, i.e. calibrated, or re-calibrated to fit real data and which further allows accurate dynamic prediction of considered circuit performance.
For using such a performance model as described above, one or more circuit parameter values of the refrigeration circuit are measured. The measuring can be done by reading in, i.e. sampling, values from at least one of the several of the temperature and pressure sensors that are dispersed throughout the refrigeration circuit. For example, sensors can be arranged on or in the compressor, the tubes, the expansion valve, etc. Also, values other than temperature and pressure values can be used instead or in addition to the above referenced circuit parameter values such as the compressor supply power, etc. The circuit parameter values can be periodically measured and then, for example, stored in a memory that is connected to the refrigeration circuit.
Advantageously, it has been shown that by interconnecting a first and a second performance model as described above allows to predict the performance and the behavior of the considered refrigeration system in operation with a better accuracy and robustness than standard pre-determined and components-specific reference models. This scheme can be referred to as cross-mapping and can be used for predicting performance, monitoring, system control and failure detection purposes.
It has been found that the discharge line temperature and the suction flow of the at least one compressor serve as very reliable indicators for performance monitoring, system control and failure detection purposes. In particular, any deviation of the flow in the actual circuit versus the performance model will affect the discharge line temperature, DLT, significantly.
Therefore, in the method according to the invention, the discharge line temperature, Tpm, of the circuit is calculated with the first performance model as a function of at least one circuit parameter value of the one or more measured circuit parameter values from the refrigeration circuit. In one example, the discharge line temperature, Tpm, could be calculated as a function of one or more of the following circuit parameter values: a number of compressors, N, running in the circuit, a suction pressure, P1, a discharge pressure, P2, and a suction temperature, T1, from the refrigeration circuit.
As an example, the discharge line temperature can be expressed as function of the same circuit parameters values by calculating an energy balance of the compressor and the system based on pre-determined compressor performance data, for example: EN12900/AHRI550 data.
In another example, the discharge line temperature can also be expressed by using a MISO, Multi-Input-Single-Output, equation structure as follows:
Tpmt=a1*Tmeast−1+a2*Tmeast−2+a3*Tmeast−3+ . . . b1*P1t+b2*P1t−1+b3*P1t−2+c1*P2t+c2*P2t−1+c3*P2t−2+d1*T1t+d2*T1t−1+d3*T1t−2
In one embodiment, all mentioned circuit parameter values are used for the calculation. A first differential value, ΔT, i.e. a relative value that defines the difference of two absolute values, can then be calculated by comparing the calculated discharge line temperature, Tpm to a measured discharge line temperature, Tmeas, from the refrigeration circuit.
Advantageously, the accuracy and the robustness of the mapping are further increased by calculating a first flow, Mpm, with the first performance model as a function of at least one measured circuit parameter value of the one or more measured circuit parameter values. For example, the first flow, Mpm, can be calculated as a function of N, P1, P2, and T1 from the refrigeration circuit with the first performance model. The obtained value is then compared to a second flow, Mevm, through the expansion valve for calculating a second differential value, ΔM.
The second flow, Mevm, through the expansion valve can be calculated by means of a second performance model for the expansion valve model as a function of at least one of the one or more measured circuit parameter values.
The second flow, Mevm, i.e. the flow through the expansion valve, can be, for example, calculated with a regression-type model, where the characteristic equation for the expansion valve can be expressed as a relationship between valve flow, valve opening and system operating conditions. For example, a relationship like {dot over (m)}=F(P1,P2,φ(%),T3) can be used, where {dot over (m)} is the flow through the expansion valve, P1 is the suction pressure, P2 is the high pressure of the system (representative of the valve inlet pressure), T3 is the liquid temperature, and φ(%) is the valve opening in degree. However, such relationship could also be of an Auto-regressive form.
The first differential value, ΔT, and the second differential value, ΔM, are then evaluated, where evaluating could be, for example, an analysis or determination whether the first and second differential values ΔT, ΔM, are still within a certain predetermined range of their respective threshold values. The evaluation results can be used for the various applications as defined in the embodiments described below.
In one embodiment, it is indicated that at least the first performance model is ready to be used for predicting performance of the refrigeration circuit based on determining that at least one of the measured discharge line temperature, Tmeas, and the first differential value, ΔT, remains stable. Here, stable can be defined as the time derivative of the temperature value or of the temperature differential value being close to 0, for example an order of magnitude 1E-3 to 1E-6 which means that the measured discharge line temperature, Tmeas, does not vary more than, for example, 0.5° Kelvin over the entire operation period.
During the first hours of operation of the refrigeration circuit, break-in effects have a major impact on circuit performance and behavior. Here, break-in refers to the period of time that is needed until all the measured values remain within their normal operational limits, i.e. until the system has reached stable operation. For example, it might take up to 24 hours until the measured discharge line temperature Tmeas remains within a certain, i.e. normal, range during normal operation. Even though compressor performance can be already predicted for monitoring, control or fault detection purposes before break-in is done, the prediction accuracy after break-in is drastically increased.
If after the initial period of operation time, stable operation is achieved, an indication can be generated to let a system operator, or a control software running in an electronic controller know that the system break-in can now be considered as ended and that at least the first performance model can now be used for predicting the performance of the refrigeration circuit. The indication can be a signal that is sent to an electronic controller, and/or a flag that is set in a control software running in the electronic controller, etc.
In one embodiment, the first performance model is calibrated in response to the indication that the first performance model is ready to be used for predicting performance.
It is known that regression and polynomial models need to be calibrated and re-calibrated to be sufficiently precise for performance prediction, fault detection, control etc. In the description, the term calibrated is used for both, the initial calibration and the subsequent calibration(s), i.e. re-calibration(s).
As described above with regards to the previous embodiment, it might take up to 24 hours until the system has reached a stable performance level. During that initial period of operation time, the model could be used for monitoring, control or fault detection purposes as described above. However, the prediction might not be very accurate. Therefore, after the system has reached stable operation, the first performance model can be calibrated for the first time to account for possible deviations due to inaccuracy of the first performance model, because of components manufacturing variability and application variability.
Calibrating the first performance model could be done by adjusting the parameters of the first performance model so that at least one of the two values ΔT, ΔM tends towards zero. Also, the calibration could be an iterative process which is done for reducing ΔT, ΔM errors.
In another embodiment, the evaluating comprises determining whether to calibrate the first performance model based on at least one of the two differential values, ΔT, ΔM. This determining can be done, for example, periodically after the first performance model was initially calibrated in response to the indication that the first performance model is ready to be used for predicting performance. However, this determining can be also done independently from an indication that the first performance model is ready to be used for predicting performance and/or independently from a first calibration. For example, determining could be done in predetermined time intervals after the refrigeration circuit was taken in operation.
Advantageously, for determining whether the calibrated first performance model is still in accordance with the actual performance of the refrigeration circuit, it has been found that the discharge line temperature and the suction flow of the at least one compressor serve as very reliable indicators. Deviation between actual measured Tmeas and calculated Tpm, and hence an increase in the absolute value of the first differential value ΔT is monitored. If the deviation is increasing with time, without overpassing a pre-determined evolution rate, deviation is considered to be due to “normal ageing”. Then, a new calibration of the first performance model is needed. This could be done by updating the first performance model with recently measured values from the refrigeration system as described above. If not, deviation could be due to a failure in the system, e.g. a faulty compressor, and a fault detection method could be activated. For example, calibrating the performance model could be necessary after the performance model was initially calibrated, if the two values that form the differential value ΔT would differ by more than 5%. The same threshold could also apply for ΔM. Advantageously, to further increase the prediction accuracy, a deviation of the other differential value ΔM can be taken into account in addition to, or alternatively to ΔT in deciding whether to calibrate the first performance model.
In one embodiment, a sensor fault is indicated based on determining that the first differential value, ΔT, is outside a predetermined range, and/or an expansion valve fault is indicated based on determining that the first and second flow values, Mpm, Mevm, used to obtain the second differential value, ΔM, differ by a predetermined percentage from each other. For example, this predetermined range could be a range of 0 to 20° Kelvin, i.e. the magnitude of the difference of Tpm and Tmeas is between 0° and 20° Kelvin. Also, for example, the magnitude of the difference of Mpm and Mevm, might differ by more than 20% from each other to indicate an expansion valve fault. For example, those determinations can be done periodically after the refrigeration circuit was started. It has been shown that when the first and second differential values are outside their respective ranges, as described above, at least one of the sensors that is adapted to measure the at least one circuit parameter value is most likely faulty. The determining whether a sensor fault and/or an expansion valve fault has occurred can be done right after the refrigeration circuit is started, or can be done periodically after the first performance model was initially calibrated to achieve a higher degree of accuracy.
In another embodiment, the evaluating comprises determining the presence of a fault based on at least one of the two differential values, ΔT, ΔM. Here, the presence of a fault can be determined if the two values that form the differential value ΔT would differ by more than 10%. The same or a similar threshold could also apply for ΔM. For increased accuracy, the evaluating could be done with a calibrated first performance model. Nevertheless, the evaluating would already give useful results when just using a non-calibrated performance model.
Also, possible faults could be better localized by a more detailed analysis of the differential values, ΔT, ΔM for example by analyzing the second differential value, ΔM. Here, it could be checked whether the second differential value, ΔM, is positive or negative. For example, if the flow, Mevm, through the expansion value which can be calculated by means of a regression-type model as described above, differs from the flow, Mpm, which is calculated by means of the first performance model, it is highly likely that a fault in the expansion valve is detected. In contrast to a fault in “the expansion valve side”, a different type of fault could be detected, if the first flow, Mpm, which is calculated from the model differs from the calculated second flow, Mevm, through the expansion valve. In this case, a compressor fault might have appeared or there might be a problem with the commissioning of the performance model.
In one embodiment, a power consumption, Ipm, of the at least one compressor is calculated with the first performance model as a function of at least one of the one or more measured circuit parameter values and the calculated power consumption, Ipm, is compared to a measured power consumption, Imeas, from the refrigeration circuit to obtain a third differential value, ΔI, and evaluating the third differential value, ΔI. Here, evaluating could be an analysis or determination whether the third differential value ΔI is still within a certain predetermined range of its respective threshold value. The evaluation result can be also used for the various applications as defined in the embodiments described above. In the description, the term power is used interchangeably for electrical power in Watts and for the electrical current in Amperes.
The power consumption, Ipm, can be calculated as a function of at least one of the one or more measured circuit parameter values. In one example, the power consumption could be calculated from the first performance model, as a function of the suction pressure P1, the discharge pressure, P2, and a compressor speed, ω. The measured power consumption value, Imeas, can be obtained by means of a current sensor installed in the electric main line, supplying the at least one compressor in the refrigeration circuit with electric energy.
Advantageously, the presence of a system fault, sensor fault or expansion valve fault could be also determined if the third differential value, ΔI, is outside a respective predetermined range which can vary depending on the type of fault. Also, advantageously, the break-in period, as described above, can be characterized directly by tracking the evolution of the compressor's power intake. Further, by taking the power consumption into consideration, higher accuracy levels can be reached, because compressor losses can be directly estimated by comparing power measurement and refrigerant-side compression work. This allows identifying a relationship between compressor losses. Also, to further increase the prediction accuracy whether or not the first performance model needs to be calibrated, or re-calibrated the third differential value can be taken in account in addition to, or alternatively to ΔT and/or ΔM.
In yet another embodiment, a further discharge line temperature, T′pm, is calculated with the first performance model as a function of at least one or more ideal circuit parameter values of a desired operating condition and the calculated further discharge line temperature, T′pm, is compared to the measured discharge line temperature, Tmeas, from the refrigeration circuit to obtain a further first differential value, ΔT′, a further first flow, M′pm, is calculated with the first performance model as a function of at least one or more ideal circuit parameter values of the desired operating condition, a further second flow, M′evm, is calculated through the expansion valve with the second performance model for the expansion valve as a function of at least one or more ideal circuit parameter values of the desired operating condition, the further first flow, M′pm, is compared to the further second flow, M′evm, to obtain a further second differential value, ΔM′, and the opening of the expansion valve is adjusted based on one of at least the further first differential value, ΔT′, and the further second differential value, ΔM′.
Advantageously, a target expansion valve opening can be estimated based on the cross-mapping. Here, ideal circuit parameter values of a desired operating condition, which could be expressed with known circuit parameter values that correspond to the desired operating condition, can be used to deduce the target expansion valve opening by means of the cross-mapping relationship linking compressor suction flow, valve opening and operating conditions.
Therefore, at a given operating condition, performance model cross-mapping allows to determine the actual compressor and expansion operating conditions and to compare them with desired compressor and valve operating conditions.
For example, the desired operating condition could be a desired superheat condition. Here, if the actual suction state, for example, the enthalpy or the temperature, is higher than a target suction state, for example, enthalpy or temperature, the expansion valve opening can be increased to increase compressor suction flow. If the actual suction state is lower than the target suction state, the expansion valve opening can be reduced to limit compressor suction flow. This method allows robust operation in positive high, for example more than 3° Kelvin or low, for example less than 3° Kelvin superheat operation, but also in slightly floodback operation. Here, the term floodback refers to the condition when liquid refrigerant droplets returns to the inlet of the running compressor.
Therefore, a floodback rate which can be used to refer to the refrigerant quality/state at the compressor inlet can be estimated and kept under control. Advantageously, the system can be controlled to reach a desired compressor discharge state. In this case, similar methods can be reversed to achieve this target.
However, the desired operating condition could be also a desired discharge condition. If this is the case, the system can be controlled to reach the desired discharge condition.
In another embodiment, a further power consumption, I′pm, of the at least one compressor is calculated with the first performance model as a function of at least one or more ideal circuit parameter values of the desired operating condition and the further calculated power consumption, I′pm, is compared to the measured power consumption, Imeas, from the refrigeration circuit to obtain a further third differential value, ΔI′, and the opening of the expansion valve is adjusted based on the further third differential value, ΔI′.
According to the invention, the apparatus for performance model cross-mapping in a refrigeration circuit containing at least one compressor and an expansion valve, the apparatus comprising: means for measuring one or more circuit parameter values of the refrigeration circuit, means for calculating a discharge line temperature, Tpm, with a first performance model as a function of at least one of the one or more measured circuit parameter values and means for comparing the calculated discharge line temperature, Tpm, to a measured discharge line temperature, Tmeas, from the refrigeration circuit to obtain a first differential value, ΔT, means for calculating a first flow, Mpm, with the first performance model as a function of at least one of the one or more measured circuit parameter values, means for calculating a second flow, Mevm, through the expansion valve with a second performance model for the expansion valve as a function of at least one of the at least one or more measured circuit parameter values, means for comparing the first flow, Mpm, to the second flow, Mevm, to obtain a second differential value, ΔM, and means for evaluating the first differential value, ΔT, and the second differential value, ΔM.
Advantageously, the apparatus could be a controlling device, or controller, or part of a controller. Also, the controller could be further adapted, i.e. in addition to verifying the accuracy of the first performance model, to also monitor the refrigeration circuit using the first performance model, and, thus, for detecting possible faults in the refrigeration circuit.
According to the invention, a method for detecting a present operational mode of a number of N compressors installed in a refrigeration circuit containing at least an expansion valve, and the number of N compressors, comprising: measuring one or more circuit parameter values of the refrigeration circuit, calculating for at least one of the possible operational modes, x, of the N compressors the respective suction flow value, Mpm[1 . . . x], with a first performance model as a function of at least one of the one or more measured circuit parameter values, calculating a present flow value, Mevm, through the expansion valve with a second performance model as a function of at least one of the one or more measured circuit parameter values, and comparing the at least one calculated suction flow value, Mpm[1 . . . x], to the present flow value, Mevm, to detect the present operational mode, if the values for the calculated suction flow value, Mpm[1 . . . x], and the present flow value, Mevm, are substantially equal.
The refrigeration system contains at least one compressor and at least one expansion valve, and consists in a closed-loop refrigeration circuit in which a refrigerant fluid flows and where the at least one compressor is dedicated to compress the refrigerant. Here, the refrigeration circuit contains at least one compressor. However, depending on the application, the refrigeration circuit may also contain multiple identical compressors and/or different types of fixed capacity or variable capacity compressors. Due to the system architecture, and/or communication and data transfer limitations, the amount and type of the N compressors currently running in the refrigeration system is not always available to the system operator. However, for efficiently controlling the refrigeration system, such as for example controlling the superheat value, it is important for the system operator to be aware of the compressors running in the system at any given time.
For detecting the present operational mode, i.e. the on and off states of the N compressors, e.g. compressor 1 is on and compressor 2 is off, etc., one or more circuit parameter values from the refrigeration circuit are measured. The measuring can be done by reading in, i.e. sampling, values from at least one of the many temperature and pressure sensors that are dispersed throughout the refrigeration circuit. For example, sensors can be arranged on or in the compressor(s), the tubes, the expansion valve, etc. Also, values other than temperature and pressure values can be used instead or in addition to the above referenced circuit parameter values, such as the compressor power, etc. The circuit parameter values can be periodically measured and then, for example, stored in a memory that is connected to the refrigeration circuit.
A first performance model of the refrigeration circuit is then used to calculate for at least one of the possible operational modes x of the N compressors the respective suction flow value, Mpm[1 . . . x], as a function of at least one of the one or more measured circuit parameter values. For example, this could be done by calculating the suction flow value, Mpm[1 . . . x], for each possible compressor running mode and by subsequently storing these values in a memory. For example, the memory could be the same memory in which the circuit parameter values are already stored. Alternatively, the suction flow values, Mpm[1 . . . x], could be also calculated one by one, i.e. after the suction flow value for a certain mode was calculated, e.g. Mpm[1], the value could be saved to be used for subsequent processing/comparing it to a reference value, i.e. before the next suction flow value, e.g. Mpm[1+1], for the next mode is calculated.
As explained above, a performance model could be any model capable of modeling the behavior of the refrigeration circuit. It has been found that the suction flow value, M, serves as a very reliable indicator for detecting the present operational mode of the N compressors. Therefore, in one example, for the possible operational modes x of the N compressors the respective suction flow values, Mpm[1 . . . x], can be calculated as a function of N, P1, P2, and T1 from the refrigeration circuit with the first performance model, where, x, denotes all possible operational modes of the N compressors, P1, is the suction pressure, P2, is the discharge pressure which is representative of the valve inlet pressure, and T1, is the suction temperature. The present flow value, Mevm, through the expansion valve is calculated with a second performance model, as for example, a regression-type model, as described above.
The calculated flow suction flow value, Mpm[1 . . . x] of at least one of the possible operational modes x, is then compared to the present flow value, Mevm, through the expansion valve. This can be done until the current operational mode is detected, i.e. until both flow values are substantially equal. Both flow values might be substantially equal if they differ by less than 5% to 10% from each other.
According to invention, an apparatus for detecting a present operational mode of a number of N compressors installed in a refrigeration circuit containing at least an expansion valve, and the number of N compressors, comprises: means for measuring one or more circuit parameter values of the refrigeration circuit, means for calculating for the possible operational modes, x, of the N compressors the respective suction flow value, Mpm[1 . . . x], with a first performance model as a function of at least one of the one or more measured circuit parameter values, means for calculating a present flow value, Mevm, through the expansion valve with a second performance model as a function of at least one of the one or more measured circuit parameter values, and means for comparing the calculated suction flow value, Mpm[1 . . . x], from the first performance model for each operational mode to the present flow value, Mevm, to detect the present operational mode, if the values for the calculated suction flow value, Mpm[1 . . . x], and the present flow value, Mevm, are substantially equal.
In the following, the present invention is further described by reference to the schematic illustrations shown in the figures, wherein:
a,b,c are charts showing the break-in period of an exemplarily compressor(s) that can be used with various embodiments of the invention;
In the shown refrigeration circuit, various sensors are mounted in various locations to sense the values of the corresponding circuit parameters. In the here shown example, the sensors are transducers that convert the circuits parameters, i.e. the physical values in electric signals, so that they can be for example supplied to an controller that can use the sensed values for further processing. As shown in
These circuit parameters can be used to describe the current operation conditions of the refrigeration circuit. Therefore, a controller, or processing unit (not shown) could be connected to the sensors, so that the circuit parameters could be used for controlling the refrigeration circuit, or for detecting sub-optimal working conditions. This is usually done by using performance models, i.e. this can be done by modeling or mapping.
As shown in
In addition to evaluating the first and the second differential values, ΔT, ΔM, for cross-mapping, the third differential value, ΔI, can be also evaluated to obtain even more accurate results.
a,b,c show the break-in period of an exemplarily compressor(s), for example an compressor as shown in
In the here shown example, break-in can be considered as accomplished after 3 to 4 hours of operation.
In the shown example, the break-in can be considered as accomplished for a compressor when one or both of the following conditions are satisfied:
In
In the here shown embodiment, a determination is made whether or not to calibrate the first and/or the second performance model based on evaluating the differential values ΔT, ΔM. For example, this determining could be done periodically after at least the first performance model was calibrated based on an indication that the first performance model is ready to be used for predicting performance.
In
The calibrating could be done by calibrating the compressor performance/efficiency curves that are comprised within the performance model(s) based on the measured discharge temperature, Tmeas. This allows more accurate prediction of current compressor performance (e.g. power consumption, suction flow, isentropic efficiency . . . ) in actual operating conditions.
For example, the calibration can be done following a deterministic approach or a recursive approach:
Afterwards, the calibrated first performance model can then be used to allow calibrating the second performance model and identifying the actual flow-operating condition relationship.
These relationships can be observed based on compressor manufacturer quality, reliability and performance data for various types of compressors and different operating conditions. The existence of such relationships may be used to calibrate, e.g. to adjust, the parameters of the first performance model, e.g. AHRI coefficients, as function of actual discharge temperature measured during compressor operation, as it can be done during calibrating the first and/or the second performance model as, for example, described in
Here, a further discharge line temperature, T′pm, is calculated with the first performance model as a function of at least one or more ideal circuit parameter values of a desired superheat operating condition, SHset. The calculated further discharge line temperature, T′pm, is then compared to the measured discharge line temperature, Tmeas, from the refrigeration circuit to obtain a further first differential value, ΔT′. Also, a further first flow, M′pm, is calculated with the first performance model as a function of at least one or more ideal circuit parameter values of the desired superheat operating condition, SHset. Then, a further second flow, M′evm, is calculated through the expansion valve with the second performance model for the expansion valve as a function of at least one or more ideal circuit parameter values of the superheat operating condition, SHset, the further first flow, M′pm, is compared to the further second flow, M′evm, to obtain a further second differential value, ΔM′, and the opening of the expansion valve is adjusted based on one of at least the further first differential value, ΔT′, and the further second differential value, ΔM′.
In addition, or alternatively to the superheat control as described above,
Therefore, a reference performance level can be computed/generated with the first performance model, where the system should be able to achieve this level of performance during its whole lifetime.
Such reference performance levels can be continuously generated and stored to allow continuous performance predicting of the refrigeration circuit
If significant deviation is detected between actual performance level and reference performance level established by the first performance model, an alarm can be raised and specific fault detection methods could be activated to identify the source of the discrepancy.
For detecting the present operational mode, i.e. the on and off states of the N compressors, e.g. compressor 1 is on and compressor 2 is off, etc. the first performance model of the refrigeration circuit is used to calculate for at least one of the possible operational modes x of the N compressors the respective suction flow value, Mpm[1 . . . x], as a function of at least one of the one or more measured circuit parameter values. Here, the suction flow values, Mpm[1 . . . x], for all possible compressor running scenario could be calculated and then stored for comparing the suction flow values, Mpm[1 . . . x], one by one to a present flow value, Mevm, through the expansion valve which is calculated with a second performance model, as for example, a regression-type model as described above. This can be done until the current operational mode is detected, i.e. until both flow values are substantially equal. Both flow values might be substantially equal if they differ by less than 2% from each other.
However, alternatively, each suction flow value, Mpm[1 . . . x], for each compressor running scenario could be calculated individually and then being compared to the present flow value, Mevm, through the expansion valve. This could be also done until the current operational mode is detected.
| Number | Date | Country | Kind |
|---|---|---|---|
| 15173519.8 | Jun 2015 | EP | regional |
