ELECTRONIC CONTROL DEVICE FOR A COMPONENT OF COMPRESSED AIR GENERATION, COMPRESSED AIR TREATMENT, COMPRESSED AIR STORAGE AND / OR COMPRESSED AIR DISTRIBUTION

Description

[0001] The invention relates to a method for controlling, regulating, diagnosing and / or monitoring a component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution, as well as an electronic control device for a component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution, which is configured to carry out the aforementioned method, according to the features of claim 1 or the features of claim 32.

[0002] Methods for controlling an entire compressor system are already known from the prior art (see WO 2010 / 072803 and DE 198 26 169 A1). In these systems, a central control unit controls and monitors a large number of components for compressed air generation and / or treatment, sometimes requiring a significant flow of data to and from the control unit.

[0003] Finally, DE 10 2011 079 732 A1 discloses a method and a device for controlling or regulating several pumps in a gas pipeline.

[0004] In contrast to the first-mentioned prior art, the present invention is based on the objective of providing a method for controlling, regulating, diagnosing and / or monitoring in connection with compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution, in which a central control device provided in the prior art can be relieved of the burden or in which the control, monitoring, evaluation, diagnosis, etc. is possible even more precisely.

[0005] This problem is solved according to the invention by a method for controlling, regulating, diagnosing and / or monitoring a component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution according to the features of claim 1, or with an electronic control device according to the features of claim 32.

[0006] Beneficial further training opportunities are listed in the sub-requirements.

[0007] A method for controlling, regulating, diagnosing, and / or monitoring a component of compressed air generation, compressed air treatment, compressed air storage, and / or compressed air distribution is proposed, wherein the component interacts with an electronic control unit, wherein the electronic control unit, for determining, replicating, or evaluating operationally relevant data, relies on one or more models that, as component-related models, contain information relevant to the structure or behavior of the component, and based on the models, as the evaluation purpose in a specific evaluation routine, either a Control, regulation, diagnosis and / or monitoring of the component or the determination, provision, prediction or optimization of operational data, operating states, operating modes, operational behavior and / or operational effects, performs this action, whereby current or historical structural information, operating data, operating states and / or measurement / sensor values ​​of the component are at least partially available in the electronic control unit as initial values, and wherein an adaptation of the model structure is carried out. through manual input, especially at the electronic control unit, through the transfer of configuration and parameter data sets to the electronic control unit, self-learning through simulations based on iteratively adapted models and / or based on a P&ID diagram of the component stored in the electronic control unit The P&ID diagram is preferably implemented as a machine-evaluable, structural model, e.g., as a graph or as a netlist.

[0008] According to a preferred embodiment, the method can further be designed to perform alternative evaluations with configurations of models containing possible different malfunctions or defects for the diagnosis of malfunctions and defects, wherein the respective degree of similarity between alternative evaluation results and real, current or historical measurement / sensor values ​​is used in a comparison step to identify the most probable malfunction or defect, or at least to exclude less probable or improbable causes of errors (malfunctions and defects) as a result of the comparison step.

[0009] In this preferred configuration, the typical behavior of the component indicative of a defect or malfunction is modeled and compared with the actual behavior. In particular, by comparing several alternative evaluations of model progressions for different fault causes with the actual operating behavior, a relatively more probable fault cause can be identified. Likewise, it is also possible to compare only one model progression for a specific fault cause with actual operating behavior and thereby assess whether, and how likely, a particular fault underlying the model is to occur.

[0010] According to a further preferred embodiment of the present invention, the method can further be configured to derive plausibility criteria for real measurement / sensor values ​​from structural models in order to detect malfunctions or defects and to verify compliance with these plausibility criteria for real, current or historical measurement / sensor values.

[0011] In this preferred embodiment, if the actual values ​​deviate from the values ​​considered plausible based on the models by more than a predefined threshold, this can be assessed as a malfunction or a corresponding error message can be issued. Such plausibility criteria can, in particular, include the comparison of temperatures and / or pressures at measuring points located upstream or downstream of each other in flow paths of media (compressed air, cooling air, cooling water, etc.), whereby systematic increases or decreases in temperature and / or pressure occur or are to be expected between the measuring points during the trouble-free operation of the components.

[0012] According to the invention, a method for controlling, regulating, diagnosing and / or monitoring a component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution is proposed, wherein the component interacts with an electronic control system, wherein models containing relevant information for the structure or behavior of the component are used to determine, replicate or evaluate operationally relevant data, and current or historical structural information, operating data, operating states and / or measurement / sensor values ​​of the component, at least partially available in the electronic control device, are used as initial values.In one possible configuration, alternative evaluations can be performed to diagnose malfunctions and defects using configurations of models that contain potentially different malfunctions or defects, whereby the respective degree of similarity between alternative evaluation results and real, current or historical measurement / sensor values ​​is used in a comparison step to identify the most probable malfunction or defect, or at least to exclude less probable or improbable causes of errors (malfunctions and defects) as a result of the comparison step.

[0013] Whether combined or independent, it can be advantageous for the detection and diagnosis of malfunctions or defects to compare the temporal profiles of operating data, operating states and / or state variables of the component, obtained from evaluations for past periods or otherwise specified, in particular calculated, with real, current or historical measurement / sensor values, whereby deviations between the evaluation results and measurement / sensor values ​​can be used to conclude that malfunctions or defects are present.

[0014] Instead of the propagated comparison between several alternative evaluations, possibly also in addition to the propagated comparison between alternative evaluations, even with only one modeled course, assuming the presence of a malfunction or defect, a conclusion can be drawn about the presence of a defect or malfunction, possibly with a qualitative or quantitative indication of the probability of the presence of a malfunction, for comparison with the actual operating course.

[0015] In another possible configuration, plausibility criteria for real measurement / sensor values ​​can be derived from structural models to detect malfunctions or defects, and compliance with these plausibility criteria for real, current or historical measurement / sensor values ​​can be verified.

[0016] The electronic control device can be fully implemented within the respective component and can be used independently of the component or in conjunction with another control device, for example a control device of an entire plant or a control device located even further externally.

[0017] The invention can be realized if the proposed control device is provided independently on a component or if the proposed method is carried out. Alternatively, the invention is also realized if the electronic control device acts on the component or if the proposed method is carried out when the component is integrated into a complete system and possibly also communicates with a control device of the complete system. In this latter implementation, the electronic control device of the component may, under certain circumstances, be fully or almost fully implemented in the control device of the complete system.

[0018] For the purposes of this application, the term "control unit of the entire system" or "central control unit" can refer, for example, to (only) a higher-level control unit for the entire system of interconnected components for compressed air generation, compressed air treatment, compressed air storage, and / or compressed air distribution, or even a higher-level control unit. Such a higher-level control unit may be implemented—even partially—in an external data center and / or in the cloud.

[0019] In this sense, the present invention solves the problem set out at the outset, namely to relieve a central control device provided in the prior art, regardless of whether it is a higher-level control device of the plant or an even higher-level control device, for example a higher-level (control) system.

[0020] In the following, a component of compressed air generation, treatment, storage, and / or distribution refers to a single device, such as a compressor, air filter, dryer, or storage tank. More generally, a component can also be understood as a unit consisting of at least two individual devices, for example, a compressed air generation and purification unit comprising a compressor and an air filter. Finally, a component can also be a part of a device, such as a part of a compressor. In a multi-stage compressor, for instance, even a single compressor stage can be considered a component.

[0021] A component model is a simplified, idealized, or approximate representation of the real-world behavior and / or structure of a component. Both the terms "structure" and "behavior" are to be interpreted broadly here. "Structure" can refer to the purely physical layout of a component. However, it can also encompass, for example, the circuitry or hierarchy of individual subcomponents. "Behavior" of the component also includes its behavior under external influences, such as electricity consumption expressed in market prices. This might be seen in a comparison between two situations where the component's inherent technical behavior is the same, but the market prices for the required electricity differ between the first and second situations.

[0022] The term "models" also includes sub-models. As mentioned earlier, the term "components" also encompasses subcomponents. When it comes to controlling, regulating, diagnosing, and / or monitoring a subcomponent, it may be sufficient to use a subcomponent model that is representative for the evaluation purpose and for the subcomponent itself. However, the subcomponent model can also be more comprehensive, i.e., represent more than just one subcomponent, or more specific, i.e., describe less than one subcomponent.

[0023] In a specifically preferred embodiment, the electronic control device makes different configurations of the component models or sub-component models and / or the type, number, sequence and / or scenarios of the evaluation, depending on the evaluation purpose.

[0024] In a further possible, optional configuration, the component model or the subcomponent model is adapted by parameterization or configuration to the properties and / or operating parameters of the specific subcomponent(s) to be considered in the respective evaluation, whereby this adaptation can be carried out manually, semi-automatically or automatically.

[0025] Adaptation is necessary, or at least advisable, for example, when operating parameters and / or properties of components and / or subcomponents have changed and / or are initially only approximately known. For automatic adaptation, models (the one to be adapted or another one(s)) can be used, particularly by iteratively and adaptively applying models to determine the properties and / or operating parameters to be updated in such a way as to achieve the best possible match between model behavior and the actually observed behavior and / or the behavior assumed for the future.

[0026] According to a further advantageous aspect of the present invention, the evaluation process also incorporates and / or derives operating data, operating states, and / or state variables of the component based on the models, for which measurement / sensor values ​​are not yet available. In particular, this embodiment utilizes a "virtual sensor," meaning that operating data, operating states, and / or state variables are derived or provided using models, without requiring a physical sensor to sample the actual conditions. This approach makes it possible to eliminate the need for sensors and to access values ​​for which sensors are not available or would be disproportionately expensive to implement.

[0027] In a further embodiment of the method, it is provided that different initial values ​​are used and / or different initialization times are selected, depending on the evaluation purpose. In a possible embodiment of the present invention, it is provided that the evaluation process takes place during the operation of the component, in particular in direct interaction, i.e., taking into account the current operating behavior of the component before, during, or after the corresponding operating behavior of the component.

[0028] In one possible embodiment of the present invention, the evaluations consist wholly or partly of the analysis of models, in particular the analysis of logical models. The component models can, for example, be physical, logical, structural, stochastic, monetary, empirical, value-based, and / or combined models of these categories.

[0029] As mentioned at the outset, the electronic control device can be integrated, at least partially, and in particular completely, into the compressed air generation, treatment, storage, and / or distribution components. However, it is also possible that it is not integrated, at least partially, into the compressed air generation, treatment, storage, and / or distribution components. For example, the electronic control device may be implemented, at least partially, in a higher-level control system, such as a system that comprehensively controls interconnected components of the compressed air generation, treatment, storage, and / or distribution system, and / or in a higher-level (control) system. Individual or all computing units or...Evaluation steps can also be carried out externally in a data center and / or in a cloud.

[0030] In one possible configuration, the electronic control unit can also consist of multiple electronic control units interconnected by data exchange. In one specific configuration, the models are implemented exclusively within the electronic control unit itself. However, it is also possible for the models to be distributed across multiple electronic control units interconnected by data exchange.

[0031] The execution, processing, and / or use of the evaluations can be implemented within the electronic control unit itself. However, it is also possible to distribute the execution, processing, and / or use of the evaluations across multiple electronic control units connected via data exchange, particularly in such a way that a given process occurs through the interaction of several interconnected electronic control units via data exchange.

[0032] In one possible configuration, the evaluations consist wholly or partly of the analysis of models, in particular the analysis of structural models.

[0033] In one possible embodiment of the present invention, the evaluation routines that can be implemented include the execution of simulations by calculating or estimating the temporal development of operating data, operating states and / or state variables of the component, in particular by numerical time integration of model equations.

[0034] In one possible configuration, the operating data used and / or derived during the evaluation include operating states and / or state variables of the components for which sensor values ​​are not yet available, in the maintenance, wear, or aging state of the component, and / or state variables for which current values ​​are not measurable or only measurable to a limited extent, and / or whose values ​​depend on the entire temporal history of the component's operation since the last maintenance or repair, and / or state variables that can only be directly recorded, especially measured, imprecisely, costly, and / or prone to errors. Therefore, the "virtual sensor" concept allows for a wide variety of evaluations without the need to physically record the actual state variables using a concrete sensor.

[0035] In one possible configuration, the models can be configured by adapting the model structure depending on the subcomponents that may (optionally) be included in the component or are in operation, and this includes parameterization. Depending on the evaluation purpose, subcomponents of the component can either be completely disregarded during model configuration, represented in a highly simplified submodel, or considered in another submodel or in the model to be configured through parameterization.

[0036] In a specific implementation, the models can be configured by adapting the model structure depending on the subcomponents that are optionally included in the component or are operational, whereby the adaptation of the model structure includes, in particular, parameterization. As mentioned at the outset, one or more subcomponent models can be representative of a respective subcomponent. However, it is also possible for submodels to be broader or narrower in scope, i.e., to go beyond the description of a subcomponent or to represent less than the structure and / or behavior of a subcomponent. In a specific implementation, however, it may be provided that the configuration of the models follows from the linking of submodels to which subcomponents are assigned, which are always and / or optionally included in the component or are operational.

[0037] The configuration of physical models can preferably be based on structural models, in particular taking into account P&I diagrams or on the basis of the P&I diagrams.

[0038] It is also possible that the P&ID diagrams may be adapted or reconfigured depending on the subcomponents that may be included (optionally) in the component or are in operation.

[0039] In one possible configuration, the results of the evaluations carried out with one or more models are used to initialize and / or as predefined influencing factors for evaluations with further models.

[0040] Preferably, the configuration of the type, number, sequence, and / or scenarios of the evaluation includes the simultaneous or sequential execution of multiple evaluations for alternative future trajectories of predefined influencing factors, in particular control commands for changing the operating mode or operating state, wherein, as a result of an evaluation of the evaluation results, a selection of the most favorable trajectories of predefined influencing factors is made. This selection of the most favorable trajectories can follow a single-stage or multi-stage process. In particular, it is conceivable to exclude relatively unfavorable trajectories in a preliminary selection and to include only relatively favorable trajectories in a final selection.In a specific embodiment of the present invention, the evaluation of the results and the selection of the most favorable future trends of given influencing variables are carried out using at least one objective function that includes one or more of the following criteria: . Energy consumption, energy costs, maximum value of electrical power consumption, number of operating state changes, usable waste heat quantity and / or temperature level of the waste heat, maintenance costs caused proportionally in the simulation horizon, pressure dew point, pressure quality.

[0041] In one specific possible design, it is envisaged that the control and / or regulation of the component includes the implementation of the selected most favorable progressions of given influencing factors.

[0042] In a preferred embodiment, the temporal profiles of operating data, operating states, and / or state variables of the component, obtained from evaluations of past periods or otherwise specified, in particular calculated, are compared with actual, current, or historical measurement / sensor values. Deviations between evaluation results and measurement / sensor values ​​are used to detect and diagnose malfunctions or defects. In this way, a highly universal and simultaneously reliable detection and diagnosis of malfunctions or defects of the component is possible.

[0043] In one embodiment of the present invention, the evaluation routines are initialized, evaluated, and used in an event-driven manner. Generally speaking, the evaluation routines are thus triggered or executed in an event-driven manner, whereby not every individual step may be event-driven; for example, a previously given initialization may be used, but the evaluation routine is then executed, evaluated, or used in an event-driven manner. Likewise, it is possible to evaluate previously executed evaluation routines only in an event-driven manner (based on specific criteria), etc. Reasons for executing evaluation routines in an event-driven manner can include, but are not limited to: changes in predefined influencing variables, operating states, and / or operating modes of the component, or a self-initiated, requested, or otherwise triggered diagnosis.Other examples of such events include a malfunction or defect occurring, a request from a control system, a user requesting an evaluation via the display, the operator requesting heat (utilizing heat recovery), .

[0044] In another configuration, the evaluations can also be performed, evaluated, and used cyclically, particularly when calculating positioning actions with a frequency of 1 x 10⁻³ s or less up to 1 minute, and preferably from 2 x 10⁻³ s up to 10 s. High-frequency evaluation may be considered, for example, when calculating positioning actions or monitoring the component, at least in cases where the system's response times are relatively short. Low evaluation frequencies, on the other hand, are suitable for optimization purposes. Here, a relatively low frequency is often sufficient, such as a daily, weekly, or monthly cycle.

[0045] Particularly when calculating positioning actions, a simulation horizon can be defined as needed and can range, for example, from 1 second to 15 minutes, or more specifically, from 1 minute to 5 minutes. However, when evaluating based on models, a simulation can also be terminated earlier according to certain termination criteria, without the entire simulation horizon being completed. This might occur, for example, if parameters and / or results deviate from a predefined corridor, exceed or fall below a predefined limit, and / or a desired result (e.g., exceeding / falling below a limit, adhering to a target corridor, etc.) has already been achieved.

[0046] In a preferred configuration, special evaluations can be carried out on request by a higher-level electronic control system.

[0047] In the method provided according to the invention for controlling, regulating, diagnosing, and / or monitoring a component of compressed air generation, it may be advantageously provided that, within the scope of the diagnostics and / or the control, regulating, and / or monitoring, the determination, provision, prediction, or optimization of operating data, operating states, operating modes, operating behavior, and / or operating effects also takes place. Optimization is generally understood to mean that operating data, operating states, operating modes, operating behavior, and / or operating effects are improved relative to a previously existing situation, without a desired optimal state actually being (yet) achieved.

[0048] More generally, it is pointed out that the aspects described as advantageous may require further development of both the electronic control device and the procedure, and therefore, advantageous aspects described in connection with the control device can also be transferred to the procedure or vice versa.

[0049] In a further preferred embodiment of the method, the evaluation process also incorporates and / or derives operating data, operating states, and / or state variables of the components based on the models, for which measurement / sensor values ​​are not yet available. In particular, the method according to the invention can also utilize so-called virtual sensors, i.e., operating data, operating states, and state variables that could potentially be physically measurable, but whose significance is not physically captured, but rather derived using one or more models.

[0050] In a further preferred embodiment, simulations are also carried out in the evaluation process by calculating the estimation of the temporal development of operating data, operating states and / or state variables of the components, in particular by numerical time integration of model equations.

[0051] Furthermore, the method according to the invention may provide that the results of the evaluations carried out with one or more models are used to initialize and / or as predefined influencing variables for evaluations with further models.

[0052] According to one optional aspect of the present procedure, the evaluation process is carried out during the component's operation. In particular, the evaluation process can also be performed simultaneously with the component's operation, especially if there is a direct interaction between the component's operation and the evaluation. In one possible embodiment of the present procedure, the evaluation process for a specific operating behavior of the component, based on a component model, is carried out before, during, or after the specified operating behavior. Thus, the evaluation process can precede the operating behavior, occur simultaneously with the operating behavior, or be performed afterward.

[0053] In an optional configuration, it is provided that the operating data, operating states and / or state variables of the components used and / or derived during the evaluation process are not yet available for the sensor values. the maintenance, wear or aging condition of the component, condition variables for which current values ​​are not or only partially measurable and / or whose values ​​depend on the entire temporal course of the operation of the component since the last maintenance or repair, or condition variables that can only be directly recorded imprecisely, costly and / or prone to error, in particular measurable, include.

[0054] Furthermore, it may be provided that the configuration of the type, number, sequence and / or scenarios of the evaluations includes the simultaneous or sequential execution of several evaluations for alternative future courses of given influencing factors, in particular control commands to change the operating mode or operating state, and that as a result of an evaluation of the evaluation results, a selection of the most favorable courses of given influencing factors is made.

[0055] In one possible configuration, the evaluation of the results and the selection of the most favorable future trends of given influencing factors are carried out using at least one objective function that includes one or more of the following criteria for the simulation horizon: Energy consumption, energy costs, maximum value of electrical power consumption, number of operating state changes, usable waste heat quantity and / or temperature level of the waste heat, maintenance costs caused proportionally within the simulation horizon.

[0056] In a preferred embodiment, the evaluations are performed on demand by a higher-level electronic control system. In particular, it is possible for the higher-level electronic control system to issue the request and / or also perform the evaluation. An example sequence could be as follows: Step 1: The higher-level control system issues a request for evaluation (evaluation request). Step 2: The component control system performs the evaluation.

[0057] Optional Step 3: The evaluation result is used by the component control and / or transferred to the higher-level control.

[0058] A person skilled in the art will recognize that some elements of the following description are not within the scope of the claims. To the extent that such a discrepancy exists, this disclosure is to be understood as merely supporting information and is not part of the invention. The invention is defined solely by the claims.

[0059] The invention is further explained below with regard to its features and advantages by means of a description of exemplary embodiments and with reference to the accompanying drawings. These drawings show: Fig. 1 Diagram illustrating the principle of a pre-simulation using an exemplary embodiment. Fig. 2 Diagram illustrating the principle of a parallel simulation using a second exemplary embodiment. Fig. 3 Diagram illustrating the principle of a post-simulation using a third exemplary embodiment. Fig. 4 Structure and integration of a component according to the present invention, specifically a stationary, oil-injected screw compressor. Fig. 5 A diagram illustrating the different operating states of a stationary, oil-injected screw compressor (prior art). Fig. 6 A diagram illustrating the time course of the electrical power consumption of a screw compressor. Fig. 7 Illustration of the control of a stationary, oil-injected screw compressor (prior art).Fig. 8 A diagram illustrating the fundamental relationship between utilization and the optimal distance between po and pu. Fig. 9 An embodiment of a simulation model. Figs. 10-13 Various possible embodiments of how a structural model, in particular a simulation model, can be processed in a control device. Fig. 14 A diagram illustrating pressure differences during the transition from pressure build-up to load operation and during the transition from load operation to pressure reduction. Fig. 15 A model-based approximation of the [missing information]. Fig. 14The illustrated circumstances. Figs. 16-18 show different approaches for assessing and categorizing deviations from a minimum pressure limit pmin. Fig. 19 illustrates a slider for setting an individual compromise between energy efficiency and pressure limit compliance. Fig. 20 shows a flowchart illustrating an embodiment of the present invention in which an algorithm cycle is used. Fig. 21 illustrates an embodiment of a model according to the present invention. Fig. 22 illustrates an embodiment of a simulation model. Fig. 23 illustrates an embodiment of a parallel simulation model.

[0060] The following section will explain in more detail what is meant by a model of a component and what applications are conceivable when evaluations are carried out based on a model for a component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution.

[0061] Generally speaking, a model is understood to be the simplification or simplified representation of a system, in this case specifically a component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution. 1. General Properties of a Component Model a) Simplification means that the model does not correspond to the component being evaluated in some properties / aspects. A model that corresponds to the component being evaluated in all properties / aspects is not a model of the component, but rather the component being evaluated itself. b) A model allows the evaluation of operating data, operating states, operating modes, operating behavior, and / or operating effects of a component without having to refer to the component itself for this analysis. It is essential that the properties / aspects of the component being evaluated that are relevant to the evaluation are represented with sufficient accuracy in the model. c) Since a model always represents a simplification of the component being evaluated, there cannot be "the one model" of a component. There will always be multiple models or sub-models for a single component.The model used for an evaluation depends on the evaluation task (question being asked). It is possible that the same model can be used for different evaluation tasks. For example, several models can be created for a stationary, oil-injected screw compressor: c.1. Oil circuit model c.2. Air circuit model c.3. Thermal model of the compressor motor c.4. Operating state model c.5.... A single model can also consider several of the aspects mentioned above. All of these models describe different properties / aspects of a stationary, oil-injected screw compressor. None of the models describes all properties / aspects of a stationary, oil-injected screw compressor (otherwise it would be a stationary, oil-injected screw compressor and not a model). Therefore, different models must generally be selected for different evaluations. 2.Models process different quantities in different ways. Models can be distinguished by the types of effects they describe. a) Physical models: For the use of models in component control, it is obvious that models describe physical effects that occur during the operation of the component. Examples include: a.1. Real-time control: Multiple applications of a simulation model of the component to determine quasi-optimal control actions for the optimal operation of the component with respect to one or more physical effects. -> Predictive simulation model a.2. Monitoring: Continuous execution of a simulation model of the component to compare the expected measured value profile with the actually observed measured value profile. If the expected (model) behavior and the actually observed behavior diverge significantly, this can indicate a malfunction of the component.-> Parallel simulation model a.3. Diagnosis: Multiple applications of a simulation model of the component to determine which component fault, stimulated in the model, best corresponds to measured value profiles recorded in the real component before the fault occurred. -> Post-simulation model a.4. Virtual sensors: In many cases, it is technically or economically impossible, or at least not desirable, to acquire measured values ​​in the machine that could be used advantageously for control purposes, for example. By applying a simulation model of the component, which is continuously compared with the real behavior of the component, it is possible to determine such measured values. -> also a parallel simulation model b) Monetary models: For the operator of a component, physical effects in the component are only of indirect interest (basically only when something is wrong with the component).Of immediate interest are models that process monetary quantities. The following applications for this type of model are conceivable, for example: b.1. Energy cost calculation: Based on the temporal history of the component's behavior (recorded in the past or calculated through predictive simulation), the energy costs for operating the component can be calculated from the electrical power consumption using a model for the cost of electrical energy (electricity tariff model). b.2. Maintenance cost calculation: Based on the temporal history of the component's behavior (recorded in the past or calculated through predictive simulation), the costs for maintaining the component can be calculated using a maintenance model. b.2.1. In a simple variant, the maintenance costs are calculated based on the component's operating hours. b.2.2.In a more complex variant, maintenance costs are calculated based on wear models of the maintenance parts (these take into account the physical environmental conditions under which the component is operated). b.2.3. Things get really interesting (but also very complex) when considering the simultaneous maintenance of several components (and thus travel cost savings) in machine networks. b.3. Total cost-optimized operation of the component: In terms of cost optimization, it would be advantageous to control and maintain the component in such a way that the total costs of the component are minimized. One could imagine that degrees of freedom exist in a component for its control, allowing for low-wear or energy-efficient operation. For stationary, oil-injected screw compressors with heat recovery, for example, the VET (VET = compressor discharge temperature) could be such a degree of freedom.The higher the VET, the greater the potential heat recovery through heat recovery. The higher the VET, the faster, for example, the oil could degrade. By offsetting heat recovery gains against oil maintenance costs, a cost-optimal VET can be calculated. c) Weighted Models: Weighted models can be used to minimize the maintenance costs of a component. Weighted models are those that incorporate subjective valuations of physical or monetary quantities in one way or another (the weighted sound pressure level serves as an example here). Wear models for components or operating fluids are examples of weighted models. c.1. Example: Air filter in stationary, oil-injected screw compressors: c.1.1. Primitive variant: The wear level of the air filter is simply calculated based on the compressor's operating hours. Once a threshold for operating hours is reached, the filter must be changed. c.1.2.A slightly more complex variant: The wear degree of a filter is plotted on a characteristic map against volume flow and differential pressure. The characteristic map is determined empirically. c.2. Example: Oil in stationary, oil-injected screw compressors: c.2.1. Primitive variant: The wear degree of the oil is simply calculated based on the compressor's operating hours. Once a threshold value for operating hours is reached, an oil change is due. c.2.2. Complex variant: The oil wear is calculated as a function of the VET (Variable Operating Time), e.g., as the integral of VET over time. If the VET-time area exceeds a predefined threshold value, an oil change is due. d) Logical models represent the behavior of control algorithms ("control engineering equivalent to physical models"). e) Stochastic models represent situations that cannot be described deterministically. 3.Models describe different situations. Models can be distinguished by what they describe. a) A model can describe information about the structure of a component. For example, a pipe and instrumentation diagram (P&ID) can describe which subcomponents a component consists of and how these subcomponents are connected within the component. The bill of materials (BOM) of a component can also be used to determine its subcomponents. Therefore, a P&ID and / or a BOM can be used to create a model. Models containing structural information are very well suited as starting models. New models can be derived from these starting models by applying analysis algorithms. b) A model can describe information about the behavior of a component. b.1.Static models describe the behavior of a component at a given operating point, neglecting transient processes that describe how the component reached that operating point. Static models are often directly accessible for evaluation (e.g., identifying the operating point with the best specific performance). b.2. Dynamic models describe the component's behavior over time (transient behavior). These models provide measured value profiles that are not directly accessible for evaluation. The measured value profiles must first be converted into key performance indicators (KPIs) before an evaluation (based on the KPIs) is possible. 4. Application of Models Models can be applied / executed in various ways within a component's control system. How a model is applied / executed also depends on the model type. a) Models can be used to derive new models.In particular, models with structural information (e.g., P&IDs) are candidates for deriving new models. This derivation occurs when an analysis algorithm interprets the original model and generates a new model using the knowledge stored within the analysis algorithm. b) Models can be used to simulate the behavior of a component. Here, the model is used as a simulation model to calculate a possible future behavior or a behavior observed in the past as a time series of the component. An algorithm core determines which simulations are performed with the simulation model and how the simulation results are interpreted. c) Models can also be used directly for optimization. For this, an optimization procedure must exist for the type of model that can directly analyze the model with respect to a given problem.Simple characteristic curves are an example of such models (e.g., specific power versus load). d) A combined application of models is particularly interesting. d.1. A simulation model of the component is derived from a P&ID diagram (initial model). d.2. The predictive simulation model is used to determine how the component will operate under given boundary conditions. The result of the predictive simulation is the time history of measured variables in the component. d.3. Based on the time history of measured variables of the component, determined by predictive simulation, the maintenance costs for operation in the coming months are calculated using wear models for maintenance parts. 5. Three application scenarios for models (behavioral models) a) Predictive simulation: In a stationary, oil-injected screw compressor with star-delta start, the pressure at the compressor outlet should be kept above 6.5 bar and below 8 bar.Using a model (simulation model), and starting from the current state of the compressor and the pressure in the tank, and assuming a constant compressed air consumption profile, the upper cut-off pressure at which the best specific output is achieved is determined. Various alternatives are tested for this purpose; see [reference]. Fig. 1The simulation results are then used (and sensibly only determined) when the compressor is under load: if the current pressure is below the cut-off pressure with the best specific output, the compressor remains under load. If the current pressure is above the cut-off pressure with the best specific output, the compressor switches off the load. Answering the questions: What is the initial time? Present. What is the initial value of the state variables? The actual compressor state (e.g., compressor under load; internal pressure 7.2 bar; compressor motor start-up time 403 seconds; pressure in the tank 7.0 bar = pressure at the compressor outlet). Which simulation period is evaluated? 10 minutes (arbitrary choice) -> complete simulation horizon. Which time profiles of the input variables am I imprinting? Constant compressed air consumption (e.g.,(determined from the pressure gradient in the past) Which model parameters are used? ∘ (Effective) volume of the compressed air storage ∘ Delivery quantity of the compressor specified by the user as a characteristic value ∘ Load and idle power specified by the user as a characteristic value b) Parallel simulation (or "concurrent simulation") .

[0062] The moisture content of the oil in a stationary, oil-injected screw compressor is to be estimated. For this purpose, a model of the combined oil / air circuit is used. The model is executed continuously, i.e., real time and simulation time run synchronously (see figure). Fig. 2 The simulation model is started once and then continues to run indefinitely. Answering the questions:

[0063] What is the initial time point? The time at which the controller starts operating. What is the initial value of the state variables? The water mass in the oil circuit is 10% of the oil mass in the oil circuit (arbitrary setting, conservative assumption: the compressor was not run dry last time). Which simulation period is evaluated? The evaluated simulation period begins when the controller starts operating and ends when the controller stops operating (power supply disconnection). Which time profiles of the input variables am I imprinting? • Current intake temperature • Current internal pressure • Current VET • Current compressor speed. Which model parameters are used? • Oil volume in the oil circuit • Relative humidity of the intake air

[0064] Therefore, according to an optional aspect of the present invention, it is also proposed to determine the moisture content in the oil using a simulation model.

[0065] The following section discusses in more detail a model for estimating the moisture content in the oil of a stationary, oil-injected screw compressor: This model assumes a screw compressor configuration in which an electric control valve is integrated into the oil circuit. The electric control valve allows the control system to influence the cooling capacity of the system. The aim of this control is to prevent condensation within the screw compressor, accelerate condensate removal, and avoid unnecessarily high temperatures in the oil circuit.

[0066] The pressure quantities used in the model are to be understood as absolute pressure. Where gauge pressure relative to the environment is used, this is indicated by an explicit pressure difference calculation. The temperatures used in the equations are absolute temperatures (unit "Kelvin").

[0067] Drastic simplifications were made in the modeling compared to the real process (screw compressor): The air volume flow rate entering the screw compressor is equal to the air volume flow rate exiting it (relative to ambient conditions). The water content of the oil circuit changes only under load. The water mass flow rate exiting the screw compressor depends solely on the internal pressure and temperature. The dependence on the water content of the oil circuit (if water is present) or the oil circulation rate (between the oil separator tank and the compressor block) is neglected.

[0068] It goes without saying that modified models can also be used that do not make the aforementioned simplifications.

[0069] To determine the water content of the oil circuit in oil-injected screw compressors, a monolithic model of the screw compressor is created. This model represents the oil circuit of an oil-injected screw compressor in a highly simplified manner. The model serves to estimate the water mass in the oil circuit, which is unmeasurable due to the lack of sensors. For this purpose, the water flows at the system boundaries (ambient environment and compressed air network), which are also unmeasurable, are estimated based on measured values. By balancing the water flows at the system boundaries and assuming a specific initial water content in the oil circuit, the water content can be determined. Only the effects that are absolutely necessary for estimating the water content are considered. Figure 21The structure of the model is shown: The screw compressor forms the core of the model. The balancing of the water flows takes place within the screw compressor. For this purpose, the mass of water flowing into the screw compressor from the surroundings is measured. ṁ 1 with the mass of water flowing from the screw compressor into the compressed air network ṁ 2 is accounted for. The difference represents the change in the mass of water stored in the oil circuit. m H 2 O (see Formula 1). dm H 2 O dt = m ˙ 1 − m ˙ 2

[0070] The mass of water flowing into the screw compressor from the surrounding area ṁ 1 is derived from the intake air volume flow V̇ 1. The intake air volume flow rate depends on the rotational speed n of the compressor and the pressure difference between the suction and high-pressure sides of the screw block. For simplicity, it is assumed that the pressure difference is determined by measuring the internal compressor pressure. p1 and the ambient pressure p amb can be determined (see Formula 2). The exact relationship between volume flow, speed and pressure difference is compressor-specific and is approximated by a characteristic curve, which can be determined, for example, through measurements on prototype systems. V ˙ 1 = f n , p i − p amb

[0071] To calculate the intake water mass flow rate, the air volume flow rate is calculated using the Clausius-Clapeyron equation and assuming a constant specific heat of vaporization for water with the relative humidity. ρ and the intake temperature T amb calculated (see formula 3). m ˙ 1 = ρ ∗ E s 273 , 15 K ∗ e q c R w ∗ 1 273 , 15 K − 1 T amb R w ∗ T amb ∗ V ˙ 1

[0072] The formula is valid for ambient temperatures between 0 °C and 100 °C. The quantities used have the following meanings: IT (273.15 K ) = 6.1 mbar: The vapor saturation pressure at 0 °C (273.15 K) q c = 2410 kJ kg The specific heat of vaporization of water R w = 0 , 462 kJ kg ∗ K : The specific gas constant of water

[0073] The outflowing mass of water ṁ 2 is based on the maximum capacity to absorb the volume of air flowing from the screw compressor into the compressed air network. V̇ 2 determined. Here, it is assumed (as a simplifying assumption in this model) that the outgoing air volume flow rate (relative to ambient conditions) corresponds to the incoming air volume flow rate (relative to ambient conditions) (see formula 4). V ˙ 2 = p amb p i ∗ V ˙ 1

[0074] First, the water mass flow ṁ 100%, assuming a relative humidity of 100% at the temperature in the oil separator tank T i , formed (see Formula 5). Formula 5 is based on the same physical principles as Formula 3. m ˙ 100 % = E s 273 , 15 K ∗ e q c R w ∗ 1 273 , 15 K − 1 T 1 R w ∗ T i ∗ V ˙ 2

[0075] The water mass flow ṁ100% is only an auxiliary value and not the actual water mass flow rate from the screw compressor. This is because the calculation did not take into account whether water is currently being fed into the screw compressor. ṁ 1 > 0) or whether there is water in the oil circuit ( m H 2 O > 0). Non-existent water cannot be discharged from the screw compressor. Therefore, a correction must be made to the calculation if necessary (see Formula 6). Formula 6 ensures that there is never less than "no" water in the oil circuit. m ˙ 2 = m ˙ 1 , wenn m ˙ 100 % > m ˙ 1 und m H 2 O = 0 m ˙ 100 % , sonst

[0076] The model presented above requires knowledge of parameters that are not (or cannot be) measured for every type of compressor. These parameters must either be calculated from other measured parameters or simply specified. The following section explains how the values ​​of some of these parameters could be determined or defined.

[0077] It is usually not intended to measure the relative humidity of the intake air. The relative humidity should be set as a constant. For worst-case scenarios with the virtual humidity sensor, a value of 100% could be set for the relative humidity. Alternatively, it is conceivable to make the value configurable via the menu.

[0078] The compressor unit speed can be assumed to be proportional to the motor speed. For screw compressors with frequency converters (VFDs), the motor speed can be read from the VFD during operation. For compressors without a VFD, the motor speed must be estimated. A simple estimation method would be to specify the motor speed or the compressor unit speed using a control parameter. The mains frequency (50 Hz or 60 Hz), the number of pole pairs of the motor, and a gear ratio (for belt-driven systems) must be taken into account.

[0079] Not every screw compressor has a sensor to measure the internal compressor pressure. pi (Printed in the Austrian Works Council). In case that pi Since it cannot be measured directly, the (always measured) network pressure is used. p N used as an approximation. p N above a surcharge for the pressure drop across the air cooler Δ N corrected (e.g., 0.5 bar). The correction term Δ N is compressor type-dependent and can be adjusted via a control parameter.

[0080] If the compressor does not have a sensor to measure the temperature in the oil separator tank T i If present, the final compression temperature will be T ADT , provided with a discount Δ T (e.g., 5 K), to estimate T i used. The temperature drop Δ T This depends on the compressor type and can be set via a menu parameter.

[0081] Another application example is parallel simulation for monitoring purposes: A concurrent simulation model is fed with influencing variables (e.g., current operating state and state changes, ambient temperature, network pressure) and generates further state variables. These include those for which real measured values ​​exist and those for which real measured values ​​are lacking.

[0082] The available real-world measurements are compared with the corresponding values ​​from the simulation model. If certain thresholds are exceeded, deviations between the two trigger warning or fault messages, and may – depending on presets, specifications, or assessments – also lead to the component being shut down.

[0083] This allows for a non-specific response to deviations from the "normal" (e.g., the undisturbed model) behavior, whereby this deviation may be the result of a disturbance for which there is no specifically defined evaluation rule from one or more sensor signals, e.g. because the evaluation rule is unknown or the required sensors are not present in the component.

[0084] Application example: The differential pressure of a filter rises to an impermissible level due to contamination. There is no differential pressure switch or sensor for the filter. However, the rising differential pressure leads to higher power consumption and / or, via higher internal back pressure, to an increased temperature. Sensor values ​​may be available for these parameters, which can be compared with the values ​​of the simulation model. Faults at one point can be detected at another by comparing sensor and model values. This can be done specifically (higher power consumption could indicate contamination of the filter(s)...) or non-specifically ("unclear what it means, but it represents an unexpected anomaly; it's best to shut down the component and perform manual, automatic, or semi-automatic fault diagnosis.").

[0085] It is possible to react to any manifestation of "implausible" component behavior without having to know a specific manifestation of a precisely defined disturbance beforehand and implement it in a monitoring function. c) Post-simulation

[0086] In a stationary, oil-injected screw compressor, the VET monitoring system triggered after the compressor motor started, meaning the VET (compressor discharge temperature) detected that it had exceeded a previously parameterized threshold, e.g., 110 °C. A subsequent simulation will be performed to determine the cause of the VET monitoring trigger. Two potential causes are known: The electric actuator for the control valve has failed. The fan motor has failed.

[0087] To facilitate the work of the operating and maintenance personnel, it should be automatically checked whether, and if so which, of the two aforementioned causes is relevant in the present case.

[0088] The post-simulation model replicates the compressor's oil / air circuit. For diagnostic purposes, the post-simulation model is initialized with the state actually observed in the compressor before the compressor motor starts.

[0089] Starting from this initial state, the post-simulation determines the VET (Variable Temperature Efficiency) over time. A parameter can be passed to the post-simulation model to specify whether the behavior should be represented when the actuator or the fan motor fails. In this example, two different alternative scenarios are simulated. In the first post-simulation, the model is configured to represent the behavior of the combined oil-air circuit when the electric actuator of the thermostatic valve is defective (the actuator is stationary). In the second evaluation, the model is configured to represent the behavior of the combined oil-air circuit when the fan motor is defective (the fan motor is stationary).

[0090] By evaluating the model in Figure 3The two different configurations result in two different VET time profiles, as shown in the figure. Comparing the recorded actual VET profile (solid black) with the VET profile determined by model evaluation for a defective valve actuator (Scenario 1) and the VET profile also determined by model evaluation for a defective fan actuator (Scenario 2), it is found that the VET profile for a defective thermostatic valve actuator corresponds better with the actually observed VET profile than for a defective fan actuator. From this, it can be concluded that, if anything, only a defective thermostatic valve actuator is a possible cause of the two known faults. d) General considerations regarding simulations Post-simulation:

[0091] The post-simulation does not need to be performed immediately when the error being analyzed occurs. It can begin and end in the past. The start in the past is linked to an event relevant to the analysis (e.g., the compressor motor starting). In the case of post-simulation, alternative past events can be calculated. This allows for a comparison of simulated behavior with actual observed behavior. The comparison can be between a single simulated behavior and the actual observed behavior. Preferably, however, the comparison includes several alternative simulated behaviors with the actual observed behavior. In the case of post-simulation, the initial point in time refers to the past. Parallel simulation:

[0092] Real time and simulated time run synchronously, i.e., at the same speed. Initialization typically occurs once when the controller starts (or possibly reinitialized upon certain events). The evaluation usually continues for as long as the controller is running. Preliminary simulation:

[0093] In the case of predictive simulation, a possible future is calculated. There is no equivalent in reality (since the real future has not yet occurred). In predictive simulation, the initial point in time refers to the real present. Generally:

[0094] The simulation is always performed in the present. In the case of forecasting and post-simulation, real time and simulated time (relative to real time) run at different rates. Simulated time (relative to real time) runs significantly faster. This is essential for forecasting. In post-simulation, this will generally also be the case, due to the high computing power available today. However, it is also possible that simulated time runs slower or at the rate corresponding to the post-simulation period. 6. "Component of compressed air generation, compressed air treatment, compressed air storage and / or compressed air distribution", as well as the arrangement of the respective control(s) of the component(s)

[0095] The following section presents several further exemplary implementations in which component models are directly used to improve component control. The component to be controlled is a stationary, oil-injected screw compressor, which conveys compressed air via compressed air treatment components (here, filters and dryers) into a compressed air storage tank. This tank then supplies a compressed air network. The resulting structure of the screw compressor, which will be used as the component here, is shown in Figure 4 depicted.

[0096] A screw compressor 11, together with other components, namely a dryer 12, a filter 13, and a compressed air reservoir 14, each forms a component of a compressor system that supplies a compressed air network 15 with compressed air at a specific pressure at a transfer point 16 between the compressed air reservoir 14 and the compressed air network 15. The screw compressor 11, considered here as an example, itself comprises several mostly integrated sub-components, namely, on the inlet side, an air filter 17, a compressor 19 driven by a motor 18, an oil separator 20, a minimum pressure check valve 21, an air cooler 22, and a compressor outlet 23. The aforementioned sub-components are arranged in series with each other in the sequence mentioned, starting from a compressor inlet 24. An inlet valve 25, associated with the compressor, is also provided between the air filter 17 and the compressor 19.Finally, a bypass line 26 with a vent valve 27 with a branch point 28 upstream of the inlet valve 25 and a connection point 29 downstream of the oil separator tank 20 is provided.

[0097] The control task is to maintain the pressure p at the transfer point 16 between the compressed air reservoir 14 and the compressed air network 15 above a minimum pressure p min and below a maximum pressure p max, while minimizing the electrical energy consumption of the component (the screw compressor 11). Between the outlet of component 11 and the compressed air reservoir 14, there are – as already mentioned – components of the compressed air treatment system that cause a pressure drop, which increases the electrical energy consumption of the component (the screw compressor).

[0098] The basic operating principle of a stationary, oil-injected screw compressor can be described by the in Figure 5The illustrated operating conditions are described. The following description applies in particular to stationary, oil-injected screw compressors with star-delta starting. For stationary, oil-injected screw compressors with frequency converters, the description may only apply to a limited extent.

[0099] In the "standstill" operating state, the compressor drive is stationary, the inlet valve is closed, the vent valve is open, and the oil separator tank is depressurized; therefore, the minimum pressure check valve is closed. The screw compressor consumes no electrical power and delivers no compressed air.

[0100] The screw compressor can be switched to the "idle" operating state via the "engine start" mode. In "engine start" mode, the compressor drive is started and brought up to operating speed. The inlet valve remains closed, and the vent valve remains open. A small bore in the inlet valve allows the compressed air generated by the rotating compressor to be circulated through the oil separator and the vent valve. Due to the dimensions of the bore in the inlet valve and the cross-section of the vent valve, a pressure of approximately 1.5 bar builds up in the oil separator during "engine start" mode. Since the minimum pressure check valve only opens at approximately 4 bar, no compressed air is released into the compressed air network, but the screw compressor does consume electrical energy. The "engine start" mode typically lasts between 4 and 10 seconds. After this, the screw compressor is in the "idle" operating state.

[0101] In the "idle" operating state, the valve positions are identical to those in the "engine start" operating state. The compressor drive also continues to rotate unchanged. The pressure in the oil separator tank remains at approximately 1.5 bar. The "idle" operating state is necessary because the compressor drive is typically only permitted to start 4 to 15 times per hour (due to the thermal stress on the motor windings caused by the starting current). To ensure that compressed air can be generated at any time, the screw compressor remains in the "idle" operating state until, after the compressor drive is switched off (transition to the "standstill" operating state), an immediate "engine start" is possible without exceeding the maximum permissible number of compressor drive starts per hour. The power consumption in the "idle" operating state corresponds to approximately 20% to 30% of the power consumption in the "load" operating state.

[0102] If compressed air is required, the operating state can be reached from the "idle" state via the "pressure build-up" state to the "load" state. In the "pressure build-up" state, the inlet valve opens and the vent valve closes. Air is drawn in from the environment through the open inlet valve, which, due to the closed vent valve in the oil separator tank, causes the pressure to increase. As soon as the pressure in the oil separator tank exceeds 4 bar and is higher than the pressure downstream of the minimum pressure check valve, the minimum pressure check valve opens. The "load" state is then reached.

[0103] In the "load" operating state, the inlet valve remains open, and air flows from the screw compressor through the air cooler into the compressed air system components. As long as there is a need to generate additional compressed air, the screw compressor remains in the "load" operating state. If compressed air generation is to be stopped, the screw compressor switches from the "pressure reduction" operating state to the "idle" operating state.

[0104] In the "pressure reduction" operating state, the inlet valve is closed and the vent valve is open. The pressure in the oil separator tank decreases to approximately 1.5 bar over a period of about 15 to 30 seconds. The screw compressor then returns to the "idle" operating state.

[0105] The screw compressor remains in the "idle" operating state until the compressor drive can be stopped and immediately restarted (-> number of permissible motor starts) or until there is a renewed need for the generation of compressed air and the screw compressor switches to the "pressure build-up" operating state in order to reach the "load" operating state.

[0106] The electrical power consumption of a screw compressor varies depending on the operating condition. This power consumption can essentially be described by the no-load power and load power ratings, which are typically found in datasheets. Figure 6 shows, in stylized form, the temporal progression of electrical power consumption as a function of the operating state. In operating state 1, "Standstill," the screw compressor consumes no electrical power. In operating state 2, "Motor Start," in addition to the no-load power, the acceleration power for the compressor rotors and the rotor of the asynchronous motor is also required. In operating state 3, "Idle," the no-load power is drawn. The no-load power is typically between 20% and 30% of the load power. In operating state 4, "Pressure Build-Up," the pressure build-up power for building up pressure in the oil separator tank is drawn in addition to the no-load power. In operating state 5, "Load Run," the load power is drawn. The load power depends on the pressure at the screw compressor outlet and increases with increasing pressure at the screw compressor outlet (by approximately 6% per bar). In operating state 6, "Pressure Release," the pressure release power is drawn in addition to the no-load power.The pressure reduction is due to the pressure in the oil separator tank, which must first be reduced before only the idle power consumption is present.

[0107] The power consumption during the operating states "engine start," "pressure build-up," and "pressure release" can be interpreted, when integrated over time, as the acceleration work, pressure build-up work, and pressure release work incurred in addition to the idling work. Expressed as the time equivalent of operating the compressor at nominal pressure under "load" conditions, the following approximate values ​​result: Acceleration work: 2s * Load running power Pressure build-up work: 1s * Load running power Pressure reduction work: 3s * Load running power

[0108] The state of the art for controlling a single stationary, oil-injected screw compressor is the use of a two-point controller with hysteresis, as described in Figure 7 is shown.

[0109] If the pressure pK at the outlet of the screw compressor falls below the adjustable threshold pu, the load request is set. If the pressure pK at the outlet of the screw compressor exceeds the adjustable threshold po, ​​the load request is reset. A set load request causes the screw compressor to enter the "load running" operating state. A reset load request causes the screw compressor to exit the "load running" operating state.

[0110] The threshold values ​​pu and po must be selected to ensure compliance with the limits p min and p max for the measured value p at the compressed air reservoir. Two aspects must be considered regarding compliance with these limits: 1. Pressure drop across the compressed air preparation components, typically 0.5 bar cumulatively across all components (relevant for pressure threshold po). 2. Time delay for transitioning the compressor from the "standby" operating state to the "load" operating state, e.g., Δt = 12 s (relevant for pressure threshold pu).

[0111] If one wants to avoid a pressure drop below p min under all circumstances (ignoring the possibility of compressor failure), the pressure limit to be selected can be determined based on the known maximum compressed air consumption. V̇ max , Given a known volume of the compressed air storage V and a known ambient pressure p amb (to which the specification of the maximum consumption refers), the following calculations can be performed: p u = p min + V ˙ max V × Δ t × p amb

[0112] If the maximum pressure drop Δp DLA across the components of the compressed air preparation system is known (occurs when the screw compressor delivers its maximum delivery volume), then exceeding the pressure limit p max can be reliably avoided by setting the pressure threshold po exactly lower by the amount of the maximum pressure drop Δp DLA. p o = p max − Δ p DLA

[0113] In particular, the conservatively chosen pressure limits pu ensure, on the one hand, maximum adherence to the pressure limits. This approach allows for pressure efficiencies of 100%, where pressure efficiencies are defined as the relative proportion of time during which the pressure p K remained within the given pressure limits pu and po. However, this comes at the cost of increased electrical energy consumption for two reasons: 1. Increasing the pressure threshold pu also increases the average pressure. The higher the average pressure, the higher the electrical energy consumption during operation under load. 2. Increasing the pressure threshold pu, while keeping the pressure threshold po constant, reduces the hysteresis range. This leads to an increase in the number of operating state changes. Generally, a higher number of operating state changes also means an increase in electrical energy consumption (although there are exceptions).

[0114] The distance between po and pu determines the frequency of operating state changes. The greater the distance, the less frequently the operating state changes and the lower the additional work required for motor start, pressure build-up, and pressure release. Simultaneously, the average pressure increases, which translates into higher power consumption during "load operation."

[0115] In practice, a pressure difference of 0.5 bar between po and pu is frequently used, where pu forms the basis for calculating pu (a low pressure level is the goal). The 0.5 bar difference is a compromise that allows for satisfactory energy efficiency in compressed air generation under both low and high compressed air consumption. However, good or very good energy efficiency cannot be achieved with this approach, as a higher or lower pressure difference between po and pu leads to optimal results depending on the compressed air consumption. The fundamental relationship between utilization and the optimal pressure difference between po and pu is shown in a stylized form in Figure 8 depicted.

[0116] Starting with a screw compressor running continuously (100% load), the optimal distance between the pressure band limits (po and pu) initially increases continuously as the load decreases. This can be explained by the fact that as the distance between po and pu increases, the switching frequency decreases, thereby reducing the energy losses associated with changing operating states. The increased power consumption due to the higher average pressure during "load operation" is more than compensated for up to a certain load. Below this load, the optimal distance decreases again, as a constant or further increasing distance between the pressure band limits would no longer compensate for the additional load power.

[0117] Due to the functionality described above, there are two significant disadvantages for the current state of the art: 1. Due to the safety margin between pu and p min, which is designed for maximum compressed air consumption, lower compressed air consumption results in an unnecessarily high average pressure, leading to increased power consumption during "load operation". 2. The statically defined limits po and pu are as described in Figure 8 The values ​​shown are actually optimal for one (or two) specific workloads. However, in the case of fluctuating compressed air consumption, which is the rule rather than the exception in practice, the distance between po and pu should be adjusted to the fluctuating compressed air consumption (and thus to the fluctuating workload).

[0118] The following presents solutions that attempt to improve the operation of a stationary, oil-injected screw compressor using simulation models by applying a simulation model of a component (here the stationary oil-injected screw compressor 11 under consideration) according to Figure 4The study determines the effects of using a given pair of parameters, pu and po, ​​on the pressure quality and energy efficiency of compressed air generation, given a specific time course for compressed air consumption. It also explores how further parameters can be derived from existing control system parameters through model evaluation, which can then be processed in other models.

[0119] By applying the embodiment described below (hereinafter referred to as control example 1), the operator of the screw compressor has the option of selecting, in very fine increments, between an energy-efficient operating mode of the screw compressor and an operating mode with a high probability of compliance with given pressure limits. The operating mode can be selected manually, in which the operator uses the model stored in the compressor control system to calculate the effects on compliance with the pressure limits p min and p max and the energy efficiency of compressed air generation for pressure thresholds pu and po specified by the operator, and then defines the pressure thresholds pu and po themselves, based on the model evaluations.Alternatively, the operating mode can also be set fully automatically, in which the operator defines limit values ​​for key figures that describe the extent of the violation of pressure limits, and an optimization algorithm independently determines the pressure thresholds pu and po based on the limit values, for which the limit values ​​are met and at the same time the energy efficiency of the compressed air generation is maximized.

[0120] The basic idea for control example 1 is to use a simulation model of the component according to Figure 4 and a given compressed air consumption profile, an optimal combination of po and pu is determined, which is then used, as is known and previously described, for calculating the load requirement via a two-point controller with hysteresis.

[0121] The simulation model consists of a model of the component (here, the screw compressor 11), which also takes into account the structure and behavior of the compressed air storage tank 14. The screw compressor model also considers the behavior of the compressor control with regard to the calculation of the load requirement via a two-point controller with hysteresis and pressure limits p0 and p0. The simulation model is a physical and logical model.

[0122] The simulation model forms the component (here the stationary, oil-injected screw compressor 11). Figure 4in such a way that, based on the simulation model and given a time course of compressed air consumption and the pressure thresholds po and pu, the time course of the pressure p and the total energy consumption of the screw compressor for the simulation horizon (corresponding in this case to the time period covered by the given compressed air consumption profile) are determined (see Figure 9 ).

[0123] The simulation model can be used to calculate, for a given time course of compressed air consumption, what time courses of pressure and electrical power consumption would result when applying specific pressure thresholds po and pu in the real component (the screw compressor 11 considered here) (within the limits of the model accuracy).

[0124] The simulation model is adapted to the specific component or subcomponent through parameterization or configuration. This is done by specifying the component properties to be used for evaluating the simulation model (e.g., delivery quantity in the operating state "load run", power consumption in the operating state "idle run", characteristic curve for power consumption in the operating state "load run", ...) and the volume V of the compressed air storage tank.

[0125] The properties mentioned above could be defined either manually or by the control devices described above (internal control, higher-level control, (control) system / data center / cloud / ...). Alternatively, the properties can be learned by the control device.

[0126] This allows one to calculate to what extent the pressure limits p max and p min for p are adhered to for a given compressed air consumption profile with a specific combination of pressure thresholds po and pu, and what electrical energy is required to operate the component to cover the compressed air consumption.

[0127] The simulation model runs in the control unit 30 of the screw compressor 11 (see Figure 10 ).

[0128] However, it is also conceivable that the simulation model runs in a control system defined as an external control unit 32, which is not part of the screw compressor (see Figure 11 For the sake of completeness, it should be noted that the dryer 12 in this configuration has an independent control unit 31.

[0129] Furthermore, it is also conceivable that the simulation model runs in the control unit 30 of the screw compressor 11, which is also connected to the dryer 12 (see Figure 12 ).

[0130] Furthermore, it is conceivable that the simulation model runs in a control unit 33 which is simultaneously assigned to the screw compressor and the dryer (see Figure 13 In this embodiment, a unit consisting of screw compressor 11 and dryer 12 forms the component within the scope of the present invention.

[0131] It goes without saying that combinations of the various methods based on the Figures 10 to 13described embodiments may exist, for example, an external control device 32 may interact with a control device 30 associated with the screw compressor, or an external control device 32 may interact with a control device 33 that controls, regulates and / or monitors a component consisting of a screw compressor 11 and a dryer 12.

[0132] The simulation model is evaluated with the aim of optimizing the operational behavior. The optimization is performed with the goal of determining the pair po and pu such that: 1. one for 1. The operator achieves sufficiently good compliance with the pressure limit p min, 2. the operator achieves sufficiently good compliance with the pressure limit p max, and 3. the electrical energy consumption of the compressor is minimized.

[0133] p min and p max are influencing factors that the operator of the screw compressor communicates to the compressor control, formed by a control device 30, 31, 32 and / or 33 or a combination of these control devices (e.g. as a manually entered parameter or message via a communication interface).

[0134] To optimize the pressure thresholds po and pu, the time course of compressed air consumption must be specified, as described above. This time course could, for example, have been determined in the compressor system itself on the previous day or week. In the simplest case, this is done by directly measuring the compressed air consumption at measuring point p, which, however, is rarely the case in practice. More often, one will attempt to estimate the time course of compressed air consumption from the pressure pK observed at the outlet of the screw compressor.

[0135] The effective storage volume V (volume of the compressed air storage tank and the piping network) is assumed to be given and constant. The current delivery rate FAD of the screw compressor is calculated from the operating state of the screw compressor and the pressure p K: If the screw compressor is operating under load, the delivery rate FAD can be determined using a characteristic curve (delivery rate FAD vs. pressure PK). If the screw compressor is not operating under load, the compressor's delivery rate is 0.

[0136] The compressed air consumption DLV can now be determined from the delivery quantity FAD using the pressure p as follows: DLV = V p amb ∗ dp dt − FAD

[0137] P amb denotes the ambient pressure as an absolute pressure to which the delivery rate FAD and compressed air consumption DLV refer. Provided the delivery rate FAD of the screw compressor does not change, the compressed air consumption is linear to the pressure gradient p. Since the pressure p is not always measured because it is a pressure outside the component, the pressure p K is used instead. Under the practical assumption that the pressure drop across the compressed air treatment components depends solely on the delivery rate of the compressors, apart from contamination effects that are only noticeable over very long periods (months), the following can be determined for the operating condition "load operation" (with an approximately constant FAD): p K = const . + p or p = p K − const . accept. Therefore, the following applies: dp dt = dp K dt

[0138] Thus, the compressed air consumption can be determined from pK as follows: DLV = V p amb ∗ dp K dt − FAD

[0139] The problem remains that around the time of the change in operating state from "pressure build-up" to "load operation" or from "load operation" to "pressure reduction," a very rapid pressure change occurs at pK, since a pressure difference between pK and p builds up during the change from "pressure build-up" to "load operation" due to the commencement of compressed air supply, which is then reduced again when switching from the operating state "load operation" to the operating state "pressure reduction" (see Figure 14 ).

[0140] The build-up and release of the pressure differential leads to inaccurate estimates of compressed air consumption. This problem can be solved by suspending the compressed air consumption estimate for a few seconds (e.g., by holding the DLV value constant) when switching to or from the "load run" operating state.

[0141] By observing the pressure profile pK during the transition from "pressure build-up" to "load operation" and vice versa, the pressure difference Δp between the measuring points pK and pK can also be estimated without measuring pK itself. As a first approximation, the pressure difference Δp between pK and pK corresponds to the jump in pK before and after the change in operating state (see [reference]). Figure 15 ).

[0142] Knowing the pressure difference Δp allows us to regulate the pressure p without measuring it.

[0143] The compressed air consumption profile determined by evaluating p K is saved as a time series in order to use it later for determining new pressure thresholds po and pu.

[0144] If reliable adherence to the pressure limits p min and p max is paramount for the compressor system operator, the pressure thresholds pu and po must be defined as described previously for calculating the load requirement. Provided the assumptions made for calculating pu and po regarding the temporal behavior of the screw compressor, the effective storage volume V, the maximum expected compressed air consumption Vmax, and the absolute ambient pressure p amb are correct, a pressure quality of 100% will be achieved, except in the case of a compressor failure. However, this rarely results in energy-optimized operation.

[0145] One way to optimize the electrical energy consumption of the screw compressor is to determine the upper pressure threshold (po) not only based on the maximum expected compressed air consumption, but also taking into account the time-dependent profile of compressed air consumption. As previously described, a high po tends to be energy-efficient at low loads and a low po at high loads.

[0146] By evaluating a model of the screw compressor, which may include information about the downstream compressed air storage tank, it is possible to determine, for a given time course of compressed air consumption, the pressure quality that results for a given pair of pressure thresholds pu and po with respect to a given pair of pressure limits p min and p max, and the electrical energy input required to generate the given compressed air consumption. The model evaluation is triggered by the operator of the compressor system. a combination of pressure thresholds pu and po, ​​the time course of compressed air consumption to be taken into account for the evaluation (e.g. by specifying "last month", "last week", "last day" or by playing in a time course of compressed air consumption), and the pressure limits to be observed p min and p max The calculation is initiated by specifying the parameter. Before each evaluation, the component model, in this case the screw compressor, is initialized so that the screw compressor is in the "standstill" operating state at the start of the simulation and the pressure in the compressed air reservoir p is a value slightly lower than p. max This choice of initial state is intended to prevent a violation of the pressure limits p. min and p max This is excluded before or after the first change in load requirements. The result of the model evaluation is the pressure quality and the electrical energy consumption required to cover the compressed air consumption.

[0147] If the operator selects the previously calculated pu value for the model evaluation and a po value that is greater than pu but less than or equal to the po value, the pressure quality will be 100%. By iteratively evaluating the model with varying po values, the operator can now determine a po value that reliably maintains the specified pressure limits p min and p max and minimizes the electrical energy consumption of the screw compressor.

[0148] If a pressure quality of 100% is not an absolute requirement for operating the compressed air-supplied process, lowering the pressure threshold pu can further reduce electrical energy consumption. This saving in electrical energy is then offset by a reduced pressure quality. By iteratively evaluating the model as described above, the compressor system operator can determine a combination of pressure thresholds pu and po that minimizes electrical energy consumption while simultaneously ensuring an acceptable pressure quality (possibly less than 100%) from the operator's perspective.

[0149] Sometimes, pressure quality is too simplistic a criterion for the operator to assess compliance with the pressure limits p min and p max necessary for the compressed air-supplied process. For example, for a compressed air-supplied process, a slight drop below p min may be uncritical if it only occurs for a short time. Therefore, it is advisable to extend the model's evaluation to calculate key performance indicators (KPIs) for different aspects of compliance with the pressure limits p min and p max. The following KPIs are examples of such indicators. Frequency of pressure limit violation taking into account a tolerance for exceeding p max or falling below p min (see Figure 16 ). Minimum and maximum pressure values ​​p. Maximum time period during which p max was exceeded or p min was undershot (see Figure 17). Maximum time-pressure area encountered for exceeding p max or falling below p min (see Figure 18 ). Total time-pressure area of ​​the period below p min. Total time-pressure area of ​​the period above p max.

[0150] Therefore, individual print quality criteria can be defined, calculated and / or monitored and thus taken into account in the determination of pu and po.

[0151] Assuming that the temporal profile of compressed air consumption from the past can be (within limits) transferred to the future, or that the temporal consumption of compressed air uploaded to the control system is representative, the two-point controller can be adapted to the compressed air-supplied process using data that is located in the control system.

[0152] However, the determination of energy-optimal pressure thresholds pu and po can also be automated, as described below.

[0153] Periodically, e.g., once a day or once a week, the historically recorded compressed air consumption profile (as a historical influencing factor stored in the compressor control) is used to determine new values ​​for the pressure thresholds po and pu. This is done by repeatedly evaluating the simulation model with different configurations regarding po and pu.

[0154] In this embodiment, a single evaluation of the simulation model does not provide information about which pair of po and pu results in the lowest electrical energy consumption of the compressor while simultaneously adhering to the pressure limits. For this purpose, multiple evaluations of the simulation model with different pairs of po and pu can be performed.

[0155] A simple way to generate combinations of po and pu for testing is to divide the interval for meaningful pressure limits, which can be defined, for example, by p min and the maximum permissible pressure pn at the screw compressor outlet, into equidistant sections (e.g., 50 mbar wide). The combinations po and pu to be tested are then simply generated by listing all meaningful pairs of section boundaries. Only pairs where po is greater than pu are considered meaningful. For example, assume that p min is 7 bar, p 8 bar, and the section width is 100 mbar. Then the following combinations of po and pu would be examined: pu = 7.0 bar: po ∈ {7.1 bar; 7.2 bar; ...; 7.9 bar; 8.0 bar} → 10 combinations pu = 7.1 bar: po ∈ {7.2 bar; 7.3 bar; ...; 7.9 bar; 8.0 bar} → 9 combinations pu = 7.2 bar: po ∈ {7.3 bar; 7.4 bar; ...; 7.9 bar; 8.0 bar} → 8 combinations pu = 7.3 bar: po ∈ {7.4 bar; 7.5 bar; ...; 7.9 bar; 8.0 bar} → 7 combinations ... pu = 7.8 bar: po ∈ {7.9 bar; 8.0 bar} → 2 combinations pu = 7.9 bar: po ∈ {8.0 bar} → 1 combination

[0156] In total, 55 combinations are tested in this example. For each test, the simulation model is initialized so that the screw compressor is in the "standstill" operating state at the start of the simulation and the pressure in the compressed air reservoir p is slightly lower than p max. This choice of initial state is intended to ensure that a violation of the pressure limits p min and p max is prevented before or after the first change in the load requirement.

[0157] The evaluation of the simulation model described here fundamentally generates fictitious operating data and operating states for the screw compressor and the compressed air storage tank (time profile of the pressure p, time profile of the electrical energy consumption of the screw compressor, time profile of the operating state of the screw compressor, etc.). These operating data and operating states are inherently fictitious because the evaluation of the simulation model uses configurations for the pressure thresholds po and pu to calculate the load requirements based on the actually observed compressed air consumption. These configurations were not used in the actual compressor control system to cover the observed compressed air consumption.

[0158] During the evaluation of the simulation model, operating data is derived for which no measured values / sensor values ​​are available to the compressor control system. This includes, for example, the operating-state-dependent electrical energy consumption of the screw compressor or the pressure p.

[0159] While the simulation model is being evaluated (multiple times), the screw compressor continues to operate normally. For calculating the load requirement, the two-point controller with hysteresis uses the pressure thresholds po and pu as determined to be energy-optimal during the last optimization cycle (e.g., one day or one week prior).

[0160] To evaluate a combination of pressure thresholds pu and po, ​​the pressure quality or the key figures for assessing compliance with the pressure limits p min and p max (the individually defined, definable, or calculated pressure quality criteria) are evaluated. This evaluation is performed by comparing the results with limit values ​​for pressure quality or the key figures that the compressor system operator has stored in the control system. All combinations for which the pressure quality or the key figures violate the operator-defined limit values ​​are rejected. From the remaining combinations, the one that results in the lowest electrical energy consumption for the given time profile of compressed air consumption is selected and used in the actual two-point controller with hysteresis to calculate the load requirement.

[0161] As an alternative to specifying detailed limit values ​​for pressure quality or key performance indicators, the plant operator can also be given an abstract weighting between energy-efficient operation and operation with a high probability of pressure limit compliance. If pressure quality is considered the relevant factor for pressure limit compliance, operators could be allowed to select a minimum pressure quality to be achieved (e.g., via a slider). The slider position would be scaled between 95% and 100% of the minimum pressure quality to be achieved, as shown in Figure 19 depicted.

[0162] For the operator, determining the minimum required pressure quality involves weighing the probability of meeting the specified pressure limits against the probability of meeting them ...

[0163] The relatively simple method of generating the combinations to be tested by equidistantly dividing the relevant pressure interval has the disadvantage that pressure threshold values ​​lying between the equidistant thresholds are not tested. This leaves optimization potential untapped. By using a stochastic optimization method, such as simulated annealing, genetic optimization, differential evolution, etc., it is possible to determine the optimal combination of p0 and p0 without discretizing the pressure thresholds to be tested. However, it is then no longer predictable how many evaluations of the model will be required to find the optimal solution.

[0164] Despite the improvements over the state of the art, the methods proposed above have two disadvantages: 1. The optimization of the pressure thresholds po and pu is carried out over a period of several hours or days, and thus generally for a compressed air consumption profile that includes phases of low, medium, and high utilization of the screw compressor. As a rule, different energy-optimal pressure thresholds result for different utilization levels, as shown in Figure 8The pressure thresholds po and pu determined through optimization cannot lead to optimal energy efficiency results in every situation, but are simply the pressure thresholds that, on average, result in the best energy efficiency. 2. The optimization of the pressure thresholds po and pu is generally based on a previously observed trend in compressed air consumption (e.g., the trend in compressed air consumption over the past week). Therefore, the optimization only leads to good results if the past trend in compressed air consumption can be projected into the future.

[0165] The first of the two disadvantages described above can be countered by shortening the time course of compressed air consumption used to optimize the pressure thresholds po and pu, and instead optimizing several pressure thresholds po and pu that are applied alternatively. For example, if it is known that the time course of compressed air consumption on weekdays Monday to Friday (normal production) is significantly different than on weekdays Saturday and Sunday (no production, essentially compressed air consumption due to leaks), then it makes sense to optimize and use different pressure thresholds po and pu for weekdays Monday to Friday than for weekdays Saturday and Sunday.

[0166] Therefore, according to an advantageous aspect of the present invention, it is also provided to provide different pressure thresholds po and pu for different operating situations, in particular for different time periods.

[0167] The second of the two disadvantages described above cannot be addressed directly within the control system of the component itself, in this case, the screw compressor. If the expected future profile of compressed air consumption cannot be derived from past observations, then knowledge of the future behavior of the compressed air-consuming process is necessary to determine the optimal pressure thresholds p0 and pu for the specific consumption profile. If a forecast of the future compressed air consumption profile is not possible through past observations or external information gathering, a simulation-based calculation of the load requirement, as proposed in control example 2 below, is a viable alternative to optimizing fixed pressure thresholds.

[0168] Control example 2 is an embodiment of the present invention, wherein a cyclic algorithm uses real-time simulations (e.g., once per second) to determine whether, for the current compressed air consumption and the current operating state of the screw compressor (see Figure 2), the following applies: Figure 20 It is energetically sensible to change the load requirement. The pressure limits p min and p max specified for measuring point p are also taken into account.

[0169] In contrast to control example 1, the simulations in control example 2 serve a different purpose. In control example 1, the simulations are used to determine which pressure thresholds would have covered a past compressed air consumption profile with the lowest electrical energy input, in order to use the determined pressure thresholds po and pu for future load command calculations. The algorithm for calculating the load requirement in the simulation model and in the actual compressor control is identical (two-point controller with hysteresis). The purpose of the simulations in control example 1 is to optimize controller parameters (pressure thresholds po and pu).

[0170] In control example 2, the evaluation of the simulation model becomes an integral part of determining the load requirement itself. Different algorithms are used for calculating the load requirement in the simulation model and in the actual control of the component (here, the compressor control).

[0171] The simulation model for control example 2 is identical to the simulation model in control example 1. It is based on a physical logic model of the screw compressor, which also takes into account information about the specific connected compressed air storage tank.

[0172] The basic idea of ​​control example 2 is to start from the current situation in a compressor system with an assumed structure according to Figure 4The aim is to investigate, by performing several simulation runs on a simulation model of the compressor system, whether it is more energy-efficient to leave the load requirement unchanged for the time being or to change the load requirement. For this purpose, various pairs of pressure thresholds po and pu are examined on the simulation model. An exemplary algorithm cycle is, as already mentioned, in Figure 20 depicted.

[0173] The algorithm cycle begins by recording and storing the current state of the compressor system. The current state includes the current compressed air consumption (e.g., determined from the pressure gradient of pK), the pressure in the compressed air reservoir pK (e.g., determined from pK), and the operating state of the screw compressor. Next, pairs of pressure thresholds po and ppu to be tested are created. These pairs can be created, for example, by discretizing the interval pmin to pp, as described previously. Finally, for each pair of po and ppu, the simulation model is evaluated to verify compliance with the pressure limits pmin and pmax for a hypothetical pressure profile p and to determine the hypothetical electrical energy consumption of the screw compressor.

[0174] Before the simulation model is evaluated, it is initialized with the state of the compressor system stored immediately after the start of the algorithm cycle. The evaluation of the simulation model thus occurs within a single algorithm cycle for all pairs po and pu, always based on the same information available in the component's control system (here, the compressor control system) about the current real state of the compressor system (e.g., operating data and operating state).

[0175] A simulation is used to calculate the hypothetical temporal behavior of the compressor system for the near future under a specific configuration for po and pu. "Near future" here refers to a time span of approximately one minute to one hour. The length of this time span is determined, among other things, by the compressor system's dimensions. Results of a simulation include (at least): the fictitious time course of the pressure p, the fictitious time course of the electrical energy E, the fictitious time course of the compressor's delivery quantity FAD

[0176] The simulation results can be used to check whether the pressure specifications p min and p max at the measuring point p would be met when controlled with po and pu, and what energy consumption is required to generate which amount of pressure (-> calculation of the specific power).

[0177] If a simulation was performed for each pair of p0 and p0, the pair that yielded the best result is determined. The best result could, for example, be the simulation result where the pressure limits are not exceeded and, in addition, the lowest electrical energy consumption is observed for the simulated period. Instead of electrical energy consumption, the specific power (quotient of electrical energy consumption E and compressor delivery rate FAD at the end of the simulated period) could also be used. Other evaluation methods are conceivable. To prevent the exclusion of energetically favorable simulation results simply because the pressure limits pmin and pmax are insignificantly exceeded, it is advisable to incorporate tolerances into the pressure limit compliance test, as described in the [reference to relevant document / reference]. Figures 16, 17 and 18 will be shown as an example.

[0178] For the best determined pair po and pu, it is checked whether the current pressure pK lies within the interval pu to po. If several best pairs po and pu were determined, the first best pair po and pu (with respect to the order of evaluation in the simulation model) is used for the check. If the pressure pK is within the interval, the current state of the load request is maintained. If the pressure pK is outside the interval, the load request is inverted.

[0179] If (even after applying a tolerance) there is no pair of po and pu for which the pressure limits pmax and pmin are met, the system checks whether the pressure pK is less than pmin. If so, the load requirement is set. Otherwise, the system checks whether the pressure pK is greater than pmax. If so, the load requirement is reset. If this is also not the case, the load requirement remains unchanged.

[0180] As described above, see above. Figures 5 to 7 The operating state is influenced by the load request. If the load request is set, the screw compressor is switched to the "load run" operating state. If the load request is reset, the screw compressor is switched to the "idle" or "standstill" operating state.

[0181] Since the evaluation of the simulation model serves to calculate the load requirement, it is clearly evident that the evaluation of the simulation model is carried out while the screw compressor is being operated.

[0182] The following section will explain in general terms how a simulation model can be evaluated, particularly through time integration.

[0183] Starting from an initial point in time and an initial state, the temporal behavior of the system described by the model is calculated. This temporal behavior is calculated by determining, from the current state of the model, where the model will be in the next time step. A numerical integration method (e.g., trapezoidal method or Runge-Kutta method) is preferably used for this purpose. The simulation, i.e., the repeated application of numerical integration to progress from one time step to the next, is performed until a termination criterion is met.

[0184] The termination criterion can be, on the one hand, reaching the end of the simulation horizon, which is defined before the start of the numerical integration. The simulation horizon defines the time range that the simulation is intended to cover.

[0185] Alternatively or cumulatively, a condition defined based on quantities calculated during the simulation can also be used as a termination criterion. During the iterative execution of the numerical integration, it is checked whether the condition is met. If this is the case, the numerical integration, and thus the simulation, is terminated.

[0186] Generally, the result of a simulation is time series of quantities described by the model. Usually, the time series are processed after the simulation, for example, by calculating new time series from existing ones (e.g., determining the time course of the total power consumption by summing the power consumption time series of the individual compressors) or by calculating key figures from time series (e.g., calculating the specific power from the initial time of the simulation to the end of the simulation horizon).

[0187] In the final step, a simulation is evaluated, for example, by comparing it with other previously performed simulations. An objective function is used for the evaluation, allowing a comparison of the processed simulation results in terms of "better than" or "worse than." Often, evaluating a simulation is only meaningful after several simulations have been performed. The evaluation is conducted over the evaluation horizon. Generally, the evaluation horizon corresponds to the simulation horizon, but it can also differ.

[0188] With reference to the Figure 22 and 23The evaluation of a model, which here is designed as a simulation model, will be explained in more detail. Particularly in view of the fact that the present invention, in a preferred embodiment, proposes that the evaluation routines be initialized, executed, evaluated, and used in an event-driven manner, especially when predefined influencing variables, operating states, operating modes, and / or when malfunctions or defects of the component occur, the question arises as to the sequence in which the individual steps are performed or initialized. The following generally applies: The model initialization is event-driven (step 3a in Figs. 22 and 23 The execution of analyses of the model is event-driven (step 3c in Figs. 22 / 23 The evaluation of simulation results is event-driven (step 3d in Figs. 22 and 23 ). In the use case of parallel simulation (see below). Fig. 23 It may be advantageous to perform steps 3a and 3c only once or rarely in an event-driven manner, while performing step 3d very frequently in a cyclical manner. Its use is event-driven (see steps 5 and 6 in the...). Figs. 22 and 23 ).

[0189] It is generally pointed out that one application of the in Figure 22 The process shown can involve the calculation of a control / regulation step. However, the order of some steps can vary. Figure 22 This does not apply to parallel simulations. These can be performed using a process according to the... Figure 23 be carried out. One use case of the process according to Figure 23 For example, the virtual humidity sensor. Also in the process after Figure 23 The order of some steps can be varied.

[0190] It should be noted that the respective events that trigger the individual steps in an event-driven manner may differ for the various steps, but do not necessarily have to.

[0191] At the in Figure 23 Regarding the outlined sequence of individual steps, it should be noted that initialization is required during the first iteration, whereas in subsequent iterations, initialization is event-driven, i.e., performed only when needed. With respect to step 3c, it should be noted that a time step (the period elapsed since the last cycle) is calculated in each cycle of the model. However, it can be advantageous to divide the time step into several integration steps. Reference symbol list

[0192] 11 Screw compressor 12 Dryer 14 Compressed air reservoir 15 Compressed air network 16 Transfer point 17 Air filter 18 Motor 19 Compressor 20 Oil separator tank 21 Minimum pressure check valve 22 Air cooler 23 Compressor outlet 24 Compressor inlet 25 Inlet valve 26 Bypass line 27 Vent valve 28 Branch point 29 Connection point 30 Control unit (screw compressor) 31 Control unit (dryer) 32 External control unit 33 Control unit (screw compressor and dryer combined)

Claims

1. Method for controlling, regulating, diagnosing, and / or monitoring a component of compressed air generation, compressed air conditioning, compressed air storage, and / or compressed air distribution, wherein the component interacts with an electronic controller, wherein the electronic control unit (11) uses one or more models, which contain relevant information for the structure or the behavior of the component (12) as component-related models, for the determination, emulation, or evaluation of operationally-relevant data, and on the basis of the models, as the intended evaluation in a specific evaluation routine, performs either a - control, regulation, diagnosis, and / or monitoring of the component or - a determination, provision, prediction, or optimization of operating data, operating states, operating modes, operating behavior, and / or operating effects, wherein current or historical structural information, operating data, operating states, and / or measurement / sensor values of the component that are available in the electronic control unit are at least partially used as initial values and wherein the model structure is adapted - by manual input, in particular at the electronic control unit, - by transferring configuration and parameter data sets into the electronic control unit, - in a self-learning manner by simulations on the basis of iteratively adapted models, and / or - on the basis of a research and innovation scheme of the component stored in the electronic controller.

2. Method as claimed in claim 1, characterized in that the electronic control unit performs different configurations - of the component models or also partial component models and / or - the type, the number, the sequence, and / or the scenarios of the evaluations depending on the intended evaluation.

3. Method as claimed in claim 1 or 2, characterized in that the component model or also partial component model is adapted by parameters and / or configuration to the properties and / or operating parameters of the (partial) component(s) to be taken into consideration specifically in the respective evaluation, wherein the adaptation can in particular take place manually, semiautomatically, or automatically.

4. Method as claimed in any one of claims 1 to 3, characterized in that operating data, operating states, and / or state variables of the component are also carried along and / or derived in the evaluation process on the basis of the models, for which measured / sensor values are not available or are not yet available.

5. Method as claimed in any one of claims 1 to 4, characterized in that different initial values are used and / or different initialization times are selected depending on the intended evaluation.

6. Method as claimed in any one of claims 1 to 5, characterized in that the evaluation process takes place during the operation of the component.

7. Method as claimed in any one of claims 1 to 6, characterized in that the evaluation process is carried out for a specific behavior, in particular operating behavior of the component on the basis of a component model, chronologically before the mentioned operating behavior or during the mentioned operating behavior or subsequently to the mentioned operating behavior.

8. Method as claimed in any one of claims 1 to 7, characterized in that the evaluations consist entirely or partially of the analysis of models, in particular of the analysis of logical models.

9. Method as claimed in any one of claims 1 to 8, characterized in that the component models are present as - physical, - logical, - structural, - stochastic, - monetary, - empirical, - assessed models, and / or - models combined from these categories.

10. Method as claimed in any one of claims 1 to 9, characterized in that it is formed at least partially, but in particular also completely, as a controller integrated in the component of compressed air generation, compressed air conditioning, compressed air storage, and / or compressed air distribution.

11. Method as claimed in any one of claims 1 to 9, characterized in that it is formed at least partially not within the component of compressed air generation, compressed air conditioning, compressed air storage, and / or compressed air distribution.

12. Method as claimed in any one of claims 1 to 11, characterized in that the specific evaluation routine which can be performed comprises the performance of simulations by calculating or estimating the chronological development of operating data, operating states, and / or state variables of the component, in particular by the numerical time integration of model equations.

13. Method as claimed in any one of claims 4 to 12, characterized in that the operating data, operating states, and / or state variables of the components used and / or derived in the performance of the evaluations, for which sensor values are not available or are not yet available, comprise - the maintenance, wear, or aging state of the component, - state variables, for which current variables are not measurable or are only measurable to a restricted extent and / or the values of which depend on the overall temporal progression of the operation of the component since the last maintenance or repair, or - state variables which are directly detectable, in particular measurable, only inaccurately, in a cost-intensive manner, and / or in a manner susceptible to errors.

14. Method as claimed in any one of claims 1 to 13, characterized in that the configuration of the models is carried out by adapting the model structure depending on the partial component (optionally) contained or in operation in the component in some cases, wherein the adaptation of the model structure in particular also includes a parameterization.

15. Method as claimed in any one of claims 1 to 14, characterized in that the configuration of the models is carried out by linking partial models, to which partial components are assigned, which are contained in the component or in operation always and / or in some cases or optionally.

16. Method as claimed in any one of claims 1 to 15, characterized in that the results of the evaluations performed using one or more models are used for the initialization of and / or as predefined influencing variables for evaluations using further models.

17. Method as claimed in any one of claims 1 to 16, characterized in that the configuration of the type, the number, the sequence, and / or the scenarios of the evaluations comprises the simultaneous or sequential performance of multiple evaluations for alternative future progressions of predetermined influencing variables, in particular control commands for changing the operating mode or the operating state, and as a result of an assessment of the evaluation results, a selection of the most favorable progressions of predetermined influencing variables is performed.

18. Method as claimed in any one of claims 1 to 17, characterized in that the assessment of the evaluation results and the selection of the most favorable future progressions of predetermined influencing variables is carried out using at least one target function, which contains one or more of the following criteria: - energy consumption, energy costs, - maximum value of the electrical power consumption, - number of the operating state changes, - usable amount of waste heat and / or temperature level of the waste heat, - proportional maintenance costs caused in the simulation horizon, - pressure dewpoint, - pressure quality.

19. Method as claimed in claim 18, characterized in that the control and / or regulation of the component comprises the implementation of the selected most favorable progressions of predetermined influencing variables.

20. Method as claimed in any one of claims 1 to 19, characterized in that the temporal progressions, which are obtained or predetermined in another manner, in particular calculated, from evaluations for preceding periods of time, of operating data, operating states, and / or state variables of the component are compared to real, actual, or historic measurement / sensor values, wherein deviations between evaluation results and measurement / sensor values are used to recognize and diagnose malfunctions or defects.

21. Method as claimed in any one of claims 1 to 20, characterized in that, to diagnose malfunctions and defects, alternative evaluations are carried out using configurations of models which contain possible different malfunctions or defects, wherein the respective degree of similarity between alternative evaluation results and real, current, or historic measured / sensor values are used in a comparison step for identification of the most probable malfunction or the most probable defect or at least less probable or improbable causes of fault (malfunctions and defects) are excluded as the result of the comparison step.

22. Method as claimed in any one of claims 1 to 21, characterized in that to recognize malfunctions or defects from structural models, plausibility criteria are derived for real measured / sensor values and the compliance with these plausibility criteria is checked for real, current, or historic measured / sensor values.

23. Method as claimed in claim 22, characterized in that the plausibility criteria include in particular the comparison of temperatures and / or pressures at measurement points, which are arranged upstream or downstream from one another in flow paths of media (compressed air, cooling air, cooling water,...), wherein in the disturbance-free operation of the components, systematic increases or decreases of temperatures and / or pressures occur or are to be expected between the measurement points.

24. Method as claimed in any one of claims 1 to 23, characterized in that the evaluation routines are initialized, performed, evaluated, and used in an event-driven manner, in particular upon the change of predetermined influencing variables, operating states, and / or operating modes of the component or in case of the request for a diagnosis.

25. Method as claimed in claim 24, characterized in that the evaluations are performed, evaluated, and used in a cyclically initialized manner, in particular in the calculation of positioning actions at a frequency of 1*10-3 seconds or less to 1 minute, particularly preferably of 2*10-3 seconds to 10 seconds.

26. Method as claimed in any one of claims 1 to 25, characterized in that, in the calculation of positioning actions, the simulation horizon is preferably 1 second to 15 minutes, particularly preferably 1 minute to 5 minutes.

27. Method as claimed in any one of claims 1 to 26, characterized in that, in the scope of the diagnosis and / or for the control, regulation, and / or monitoring, a determination, provision, prediction, or optimization of operating data, operating states, operating modes, operating behavior, and / or operating effects also takes place.

28. Method as claimed in any one of claims 1 to 27, characterized in that, in the evaluation process, simulations are (also) performed by calculating or estimating the chronological development of operating data, operating states, and / or state variables of the components, in particular by the numerical time integration of model equations.

29. Method as claimed in any one of claims 1 to 28, characterized in that the results of the evaluations performed using one or more models are used for the initialization of and / or as predetermined influencing variables for evaluations using further models.

30. Method as claimed in any one of claims 1 to 29, characterized in that the configuration of the type, the number, the sequence, and / or the scenarios of the evaluations comprises the simultaneous or sequential performance of multiple evaluations for alternative future progressions of predetermined influencing variables, in particular control commands for changing the operating mode or the operating state, and that as a result of an assessment of the evaluation results, a selection of the most favorable progressions of predetermined influencing variables is performed.

31. Method as claimed in any one of claims 1 to 30, characterized in that the assessment of the evaluation results and the selection of the most favorable future progressions of predetermined influencing variables is carried out using target functions, which in particular contain a combination of two or more of the following criteria for the simulation horizon: - energy consumption, energy costs, - maximum value of the electrical power consumption, - number of the operating state changes, - usable amount of waste heat and / or temperature level of the waste heat, - proportional maintenance costs caused in the simulation horizon.

32. Electronic control unit for a component of compressed air generation, compressed air conditioning, compressed air storage, and / or compressed air distribution, which is configured to perform the method as claimed in any one of claims 1 to 31.