Project meeting of STEP at the site of the company ATS - Agro Trading & Solutions in Hardegsen (Photo: DBFZ)

Biogas

Project ID 03KB101

FlexFeed - Flexible Feeding in Biogas Processes assisted with Model-Based Process State Recognition in Pilot Scale

Demonstration biogas plant "Unterer Lindenhof" (Photo: State Institute for Agricultural engineering and Bioenergy)

Duration

  • 01.10.2014 – 31.03.2018

Contact

Fraunhofer Institute for Solar Energy Systems ISE
Heidenhofstr. 2
79110 Freiburg im Breisgau

Dr. Stefan Junne – project manager
Telephone: +49 (0)30 314 72527
E-Mail: stefan.junne‍@‍tu-berlin.de

Prof. Dr. Peter Neubauer
Telephone: +49 (0)30 314 72527
E-mail: peter.neubauer‍@‍tu-berlin.de

Dr. M. Nicolas Cruz-Bournazou – further contact person
Telephone: +49 (0)30 314 72527
E-mail: nicolas.cruz‍@‍mailbox.tu-berlin.de

Anika Bockisch – further contact person
Telephone: +49 (0)30 314 72190
E-mail: anika.bockisch‍@‍tu-berlin.de

Partner

03KB101B - Fraunhofer Institute for Solar Energy Systems ISE
03KB101C - Fraunhofer Institute for Solar Energy Systems ISE

Associated partners

Cooperation partners

Subpartner

Results

Final Report FlexFeed (2018) (only in German available)

Topic

The biological part of biogas production can be operated flexible and controlled with a feeding management due to the storage capacity of biomass. What is missing often so far is a suitable amount of data and a methodological examination for the investigation of the impact of a flexible feeding management on product synthesis, especially under consideration of process monitoring and sampling strategies in order to reduce operative risks. Therefore, the project FlexFeed aims to investigate, if a flexible feeding management supports the sustainability and a demand-driven biogas production, and if mechanism recognition as a method can support a robust and stable process under changing loading rates. In order to achieve this, parameters of a model are fitted at different process conditions. The mechanism recognition identifies the set of parameters, which can describe the actual process state the closest. Then the process state is evaluated and can be combined with a corresponding control. The big advantage lies in the application of simplified relations between measured parameters, which are weighted based on the process state. So far, signals are considered as single values. This leads to a loss of information. Lance-based sampling systems of a previous project and novel sensors for the implementation at relevant locations in the liquid phase are applied and their relevance for the described methodology is quantitatively evaluated.

Aim

The project FlexFeed aims to evaluate and optimize a strategy for the feeding management in biogas plants, which combines innovative sensor applications in the fluid phase, model-based monitoring and neural network-based prediction as methodological approaches. The goal is to identify the real process state based on mechanism recognition, and hence to discover process disturbances very early. This is especially relevant whenever the loading rate and feedstock composition is changed. Lance-based sampling systems as well as sensors, which are located at critical spots in the liquid phase, supply the early recognition of process disturbances. The application of a simulation tool at changing loading and feedstock composition with neural networks leads to a comprehensive approach based on prediction, on line measurement and model-based monitoring. Data of very flexible states with a high feedstock load support the optimization and evaluation of the developed models.

Measures

In a cooperative project between the partners University of Hohenheim, TU Berlin and SOTAsolutions, changes of the loading rate at the test biogas plant “Unterer Lindenhof” will be conducted and impacts will be investigated.

TU Berlin

The long-term trials at the test plant will be accompanied by the application of additional sensors. Therefore, measurements are performed at different depths at various spots. In the second half of the project, different model-based suggestions for the operation will be made. Parameters are adopted for different scenarios of process disturbances. The actual state will be detected based on the mechanism recognition. Sampling intervals will be suggested due to the detected process state and the corresponding parameter sensitivities for process risk reduction.

LAB

Firstly, the spatial distribution of intermediates in the fermenter based on feedstock composition, substrate quality, loading intervals and feedstock pre-processing as well as stirring will be observed. A lance-based system is applied and critical zones are detected based on biological tracer substances. Fluid flow velocities and the viscosity of substrates determined with a pipe viscometer are measured and compared with a mixing grade model.

Secondly, ingredients of feedstock are determined online with near-infrared spectroscopy. In lab trials, the application of an acoustic wave guide for the determination of process parameters is evaluated.

Finally, a distinct feeding profile will be realized at the pilot plant, which will also lead to critical process conditions. These conditions should be recognized early with the applied sensors, and a suitable control strategy will be developed based on the data gained previously.  

Sota Solutions

The company SOTA SOLUTIONS GmbH applies the data of the cooperation partners for the training of self-learning systems. The operation of the plant will be optimized on the basis of this prediction in order to identify suitable feeding intervals. Both, aspects of process stability and economic benefits will be considered for the setup of the feeding management. The simulation of the gas production is realized by using multiple blackbox and whitebox approaches.

Focus

TU Berlin

  • Installation of sensors
  • Data quality
  • Multiparametersensors
  • Parameter reduction
  • Mechanism recognition

LAB

  • Identification of critical spots in the digester
  • Installation and validation of an acoustic wave guide
  • Optimization of the developed models in industrial scale

Sota Solutions

  • Data analysis, validation and preprocessing
  • Development of predictive analytics
  • Objective functions for efficiency and process stability
  • Methods for the optimization of feeding
Installation for mobile sampling and sensor positioning in the biogas plant (Photo: State Institute of Agricultural Engineering and Bioenergy)
Sampling point at the digester of the biogas demonstration plant "Unterer Lindenhof" of the university Hohenheim (Photo: DBFZ)
Propeller incline of the shaft agitator of the demonstration plant "Unterer Lindenhof" of the university Hohenheim (Photo: DBFZ)

Back

Project Search

Topics

Further networking

enArgus
Project database