Search

EP-4424810-B1 - DIAGNOSING FAULTS DURING OPERATION OF AN ANAEROBIC DIGESTER

EP4424810B1EP 4424810 B1EP4424810 B1EP 4424810B1EP-4424810-B1

Inventors

  • DONOSO BRAVO, Andrés Eduardo
  • SADINO RIQUELME, María Constanza
  • VALDEBENITO ROLACK, Emky Héctor
  • HANSEN FERNÁNDEZ, Felipe Eduardo
  • GÓMEZ PASCUAL, Daniel

Dates

Publication Date
20260506
Application Date
20230301

Claims (13)

  1. A method for diagnosing operation faults during operation of an anaerobic digester (10), comprising: - a preparation stage, wherein a mathematical model of anaerobic digestion (MM-AD) is run in a computing system, the preparation stage comprising: - selecting (210) an operation fault to be diagnosed; - selecting (210) an operational parameter OP-P that is associated with the selected operation fault; and - running (220) multiple simulations in the MM-AD to: - select (230, 240) an output parameter OUT-P that varies with a variation of the operational parameter OP-P; and - determine (250, 260, 270, 280) a threshold deviation of the output parameter OUT-P that indicates that the operation fault is occurring; and after the preparation stage: - a digester operation stage, wherein a mathematical model of anaerobic digestion (MM-AD) is run in a computing system, in parallel with the physical operation of the anaerobic digester, to simulate the operation of the digester over time, the digester operation stage comprising: - inputting in the MM-AD nominal values of kinetic parameters and operational parameters of the digester; - starting (410, 420) the operation of the digester and the MM-AD simulation; - obtaining (430) over time measured values MV of the output parameter OUT-P of the digester, using sensors and/or by performing sample analysis; - obtaining (440) over time expected values EV of the output parameter OUT-P, as provided by the MM-AD; - comparing (450) the measured values MV and the expected values EV of the output parameter OUT-P, over time; - determining (460) that the operation fault is occurring if the deviation between the measured value MV and the expected value EV of the output parameter OUT-P is above the threshold deviation determined in the preparation stage.
  2. A method according to claim 1, wherein the preparation stage comprises selecting more than one operation fault to be diagnosed and, for each operation fault: - selecting (210) an operational parameter OP-P that is associated with the selected operation fault; and - running (220) multiple simulations in the MM-AD, to: - for each selected operational parameter OP-P, select (230, 240) an output parameter OUT-P that varies with a variation of the corresponding operational parameter OP-P; and - for each selected output parameter OUT-P, determine (250, 260, 270, 280) a threshold deviation that indicates that the selected operation fault is occurring.
  3. A method according to claim 2, further comprising, after the preparation stage: - a digester operation stage, wherein a mathematical model of anaerobic digestion (MM-AD) is run in a computing system, in parallel with the physical operation of the anaerobic digester, to simulate the operation of the digester over time, the digester operation stage comprising: - inputting in the MM-AD the nominal values of the kinetic parameters and operational parameters of the digester; - starting (410, 420) the operation of the digester and the MM-AD simulation; - obtaining (430) over time measured values MV of the output parameter OUT-P of the digester, using sensors and/or by performing sample analysis; - obtaining (440) and comparing (450) over time measured values MV and expected values EV of each selected output parameter OUT-P, and - determining (460) that one of the selected operation faults is occurring if the deviation between the measured value MV and the expected value EV of one of the selected output parameters OUT-P is above the corresponding threshold deviation determined in the preparation stage.
  4. A method according to any one of the preceding claims, wherein the preparation stage comprises running (220) multiple simulations in the MM-AD to determine threshold deviations of the output parameter OUT-P that indicate that the operation fault is occurring, for at least two different time periods.
  5. A method according to claim 4, wherein threshold deviations of the output parameter OUT-P that indicate that the operation fault is occurring, are determined for a time period of one day.
  6. A method according to claim 4, wherein threshold deviations of the output parameter OUT-P that indicate that the operation fault is occurring, are determined for a time period of several consecutive days.
  7. A method according to any one of the preceding claims, wherein selecting an output parameter OUT-P that varies with a variation of the operational parameter comprises performing a sensitivity analysis of output parameters.
  8. A method according to claim 7, wherein the sensitivity analysis comprises: - selecting several output parameters that may vary with a variation of the operational parameter; - simulating the variation of each output parameter, over a predetermined period of time, when the operational parameter varies with respect to a nominal value to be employed in the digester operation; and - comparing the variations over the predetermined period of time of the plurality of output parameters, for the same variation of the operational parameter; and - selecting at least one output parameter for which the variation over time is higher than for other output parameters.
  9. A method according to any one of the preceding claims, wherein the preparation stage comprises calibrating the kinetic parameters of the MM-AD of an anaerobic digester in which faults are to be diagnosed.
  10. A method according to any one of the preceding claims, wherein the operation stage comprises calibrating the kinetic parameters of the MM-AD of the anaerobic digester.
  11. A method according to any one of the preceding claims, wherein at least one selected output parameter is the biogas output flow of the anaerobic digester, and/or the methane output flow of the anaerobic digester.
  12. A system for diagnosing operation faults during operation of an anaerobic digester (10), the system comprising: - a computing system (70) with a mathematical model of anaerobic digestion (MM-AD), and - a processor (60) configured to perform the method of any of claims 1 to 11.
  13. An anaerobic digester system, comprising: - an anaerobic digester (10); - sensors (50), and/or a sample collection system, to obtain measurements of output parameters during operation of the anaerobic digester (60); and - a system for diagnosing operation faults during operation of the anaerobic digester, according to claim 12.

Description

The present disclosure is related to methods and systems for diagnosing faults during operation of an anaerobic digestion, e.g. faults in the anaerobic digestion process of the anaerobic digester, by employing the results of simulations of the anaerobic digestion process in a particular reversed engineering approach, using the output to detect input variations; and to anaerobic digestion systems in which these methods and systems are applied. BACKGROUND Anaerobic digestion is a widely used process in industrial and domestic waste management and production of fuels, such as biogas. Most commonly, anaerobic digesters are used for producing biogas, the digesters comprising a sealed vessel known as a reactor or digester, which may be designed and constructed in various shapes and sizes, in order to perform a substantially controlled anaerobic digestion process within. Generally, anaerobic processes may be slow and, when using a digester, they usually involve a lot of operator supervision, in order to keep the working conditions of the digester stable or within a specific range, to produce a desired amount of biogas. Even though the anaerobic digestion process itself may be unpredictable or very prone to changes in the conditions in which it happens, there are systems which help predict the outcome of a digester (such as the production of biogas). Some of such systems use simulations of the process occurring in the digester using mathematical models of the anaerobic digestion (MM-AD in the following), such as the model known as Anaerobic Digestion Model No. 1 (ADM1), or others such as BIOmodel, AMOCO, etc., which models the operation of the digester, and are able to simulate the biogas output and other parameters over time during operation of the digester. More precisely, an MM-AD can simulate the steady state and dynamic behavior of an anaerobic digester. Traditionally, such models have been used to predict the biogas production of anaerobic digesters. The diagnose of operation faults or malfunctions during operation of an anaerobic digester are currently performed only by reacting to measurements of the digester outputs on the basis of general references e.g. given in handbooks, best practice protocols, etc. Consequently, fault detection is not fast and efficient, and after a fault is diagnosed it may take several weeks before the anaerobic process returns to stable operation. ERSAHIN MUSTAFA EVREN ED - MORENO JAIME A ET AL: "Modeling the dynamic performance of full-scale anaerobic primary sludge digester using Anaerobic Digestion Model No. 1 (ADM1)", BIOPROCESS ANO BIOSYSTEMS ENGINEERING, SPRINGER, DE, vol. 41, no. 10, 12 July 2018 (2018-07-12), pages 1539-1545 describes models that can be used to predict full-scale sludge digester performance. XUE LEI ET AL: "Nonlinear model predictive control of anaerobic digestion process based on reduced ADM1 ", 2015 10TH ASIAN CONTROL CONFERENCE (ASCC), IEEE, 31 May 2015 (2015-05-31), pages 1-6 describes an algorithm to control the anaerobic digestion process in biogas plants. ORDACE A ET AL: "Predictive control of anaerobic digestion of wastewater sludge. A feasibility study", 2012 16TH INTERNATIONAL CONFERENCE ON SYSTEM THEORY, CONTROL ANO COMPUTING (2012-10-12) pages 1-7 describes analysis, modeling, identification and predictive control of an anaerobic digestion process. Patent document WO 2013/138690 A1 describes photobioreactors for cultivation and/or propagation of a photosynthetic organism. Patent document US 2009/048816 A1 describes a method for online prediction of performance of a fermentation unit. KAZEMI PEZHAM ER AL: "Data-driven techniques for fault detection in an anaerobic digestion process", PROCESS SAFETY AND ENVIRONMENTAL PROTETCTION, INSTITUTION OF CHEMICAL ENGINEERS, RUGBY, GB, vol. 146, 24 December 2020 (2020-12-24), pages 905-915 describes a framework for detecting random faults in AD processes. Thus, there is a need to detect and differentiate operation faults in anaerobic digesters in a sufficiently advanced manner over time, in order to enhance the prediction of malfunctions, thus minimizing the time of reaction and correction of said faults, maintaining a high productivity of biogas, and reducing the risk of idle times. The inventors have now found that faults can be detected and anticipated by using a reverse engineering process to harness the outputs of an MM-AD simulation and assess the input and operational conditions that have caused those outputs, by comparing them with previously predicted reference values. SUMMARY According to an aspect of the present disclosure, a method for diagnosing operation faults during operation of an anaerobic digester is provided, comprising a preparation stage, wherein a mathematical model of anaerobic digestion (MM-AD) is run in a computing system. The preparation stage comprises: selecting an operation fault to be diagnosed;selecting an operational parameter OP-P that is associated with the selected operat