CORDIS - EU research results

Wood chip feeding technology of the future for small-scale biomass boilers

Final Report Summary - BIOCHIPFEEDING (Wood chip feeding technology of the future for small-scale biomass boilers)

Executive Summary:
Presently about 0,7 million smallscale boilers are sold in Europe, an increase until 2020 is expected. Woodchipsboilers are one main technology in this capacity range, in general this technology reached a very high level. However, the fuel feeding systems are still a considerable weak point during field operation, they are about 80% responsible of unexpected shutdowns, reduced plant availability, increased emissions and reduced efficiencies. Common woodchipfeeding systems discharge from the bottom of the storage, which implies the disadvantages of massive constructions and high operations costs. The objective of the Biochipfeeding project was to develop a new feeding system for woodchip heating plants, which shall provide a more robust operation, at reduced investment and operation costs, considering lower emissions and higher efficiency. The new system is based on a gripper, which enables material withdrawal from the top of fuelpiles in the storage, and transports it to an intermediate storage container. The gripper is equipped with sensors for online screening of the fuelquality regarding particle size, moisture content and ashcontent. Moreover, a new fuel feeding screw is developed, which allows a constant transport of the woodchips from the intermediate storage into the furnace independently of shape and size of the woodchips. Finally a fully automated controle system is implemented, which uses the information about the properties of the fuel stored at different position gained from the gripper sensors and information from the boilers controle system regarding the actual operation mode to optimize the fuel quality by appropriate fuel mixing. Thereby, the material withdrawal by the gripper and the subsequent mixing in the intermediate storage is controlled in a way, that always a fuel mixture with rather constant properties, which are adjusted to the actual needs of the combustion process, is supplied to the boiler. With this new technology the boiler performance shall be significantly improved due to a more even fuel supply with a constant quality and due to the adjustment of the fuelquality to the boiler operation mode. During the first project period, firstly the boundary conditions for the new feeding system have been defined and then the detailed concept has been developed as well as the conceptual design of the new feeding technology has been performed. Besides dedicated experimental work ( e.g. regarding certain relevant woodchipproperties, sensor evaluations, functionality tests of gripper components, etc.) also computer aided tools, such as discrete element method (DEM) modelling as well as stress simulations have been applied within the development and design of single components. The construction of a first testing plant was started. In the second project period the gripper was designed, a crane as well as the sensors positioned on the gripper were chosen. The feedingalgorithm was programmed respecting the demands of the combustion process, the demands of safety and economics. Two prototype plants were installed, one at HDG in Massing, one at Bios in Graz. One of the main objectives from the tests at Bios was the verification of the expected implication to the combustion process. After the testruns at HDG in Massing the prototype was revised - mechanical as well as the controler. The optimized prototype was tested till the end of the project. The intermediate storage inclusive the stirrer and the stoker have been optimized by TU- Munich. The cost calculations have been revised, detailed costs of production, purchasing and operation are available.

Project Context and Objectives:
WP1 Definition of the boundary conditions and development of a detailed concept

T1.1 Definition of the boundary conditions

• Kick off meeting was organized by HET, location: HDG in Massing
• workshop to discuss the boundary conditions was held on the second day in Massing
• the list of requirements have been done

T1.2. Development of concepts fort he single components , basic concepts for the different subsystems

• TU Graz/ITL worked out a basic concept for the gripper design
• TU München/FML determined properties of wood chips relevant fort he design of the feeding screw. BIOS supported with data lists and experiences
• FIR and Sinte elaborated a basic concept for the overall process
• BIOS worked out with HDG and HET a concept fort he implementation of the feeding system at a HDG woodchipboiler

T1.3. Definition of the final overall concept

• meeting was organized by HET, location: FIR in Bozen. Discussion about the single concepts. Discussion about the overall concept and definition of itself

WP2 Conceptional design of the new feeding technology

T2.1. Design of the subsystems

• TU Graz/ITL worked out the detailed design of the feeding technology, especially of the gripper. The fuel screening sensors have been selected
• TU München/FML designed the screw feeder by using the discrete element method (DEM). DEM-simulation for different screw geometries
• FIR and Sinte implemented central or decentral modules fort he overall control system

T2.2 Preparation of draft construction drawings

• TU München/FML prepared the drawings fort he screw conveyor, detailed DEM simulations
• TU Graz/ITL prepared a 3D CAD model of the gripper and draft construdtion for further components
• FIR prepared a scheme of the overall system control fort eh test facility

WP3 Construction of two first prototype plants including the modification of the automatic boiler controle

T3.1. Design of the prototype plants

• Preparation of the construction drawings
• engineering of the prototype plants
• definition of the purchased parts

T3.2. Preparation of the experimental plans

• longtermmonitoring was prepared for the plant at HDG in Massing
• specific aspects of the technology: emission, measurments, full and partial load etc.
• systematic and common procedure concerning the assessment of the technology has been developed

T3.3. Construction of the prototype plants

• two plants have been constructed including the fuel storage, the new feedingsystem and the combustion system. One plant was installed at HDG and one at BIOS
• all subsystems have been implemented

T3.4. Integration of the new process control system into the prototype plants

• the complete automation technology including overall process control system, communication interfaces, servo drives, actuators, sensors... has been integrated
• Elaboration of a supervising controle strategy to handle each subsystem

T3.5. Comissioning of the prototype plants and preparation of data recording and visualisation

• the data recording system was prepared and tested
• the data visualisation system was prepared and tested
• Installation of the datarecording system and the visualisation system in both plants
• data eavaluation tools have been developped
• Identification of the needed hardware

WP4 Performance and evaluation of test runs

T4.1 Performance of test runs for fuel feeding system validation

• Performance of test runs without boiler operation
• Verification of the proper operation of the fuel handling system
• Evaluation of the gripping behaviour
• Testing of the fuel screening technology
• Identification and summarisation of the weak points as well as optimization potentials

T4.2 Performance of test runs for overall system evaluation

• Performance of test runs with boiler operation
• Evaluation of the performance of the overall system
• Measurement and recording of different parameters (e.g. temperature load, heat counters, heating cycle temperature)
• Test runs will be performed according to the experimental plan (D3.2) applying different types of wood chips with different properties in terms of moisture content and particle size distribution
• Identification and summarisation of the weak points as well as optimization potentials

T4.3 Performance of long-term tests

• long-term test with the prototype plant placed at HDG
• Evaluation of the general operation of the technology


T5.1 Definition of proposals for optimisation

• Identification and summarisation of the weak points as well as optimization potentials by the evaluation of the long-term test runs as well as of the dedicated test runs
• Decision which optimization measure will be implemented
• Documentation within a supplement book of technical specifications

T5.2 Engineering of the optimised prototypes

• Improvement of the gripper technology as well as moisture content and particle size sensors
• Further design and dimensioning steps, if required
• Further simulation and analysis of the conveying process of Feeding Step2, if required
• developments of solutions for discovered overall control system-deficiencies and/or adaptation of the process platform to meet the requirements of possibly changed subsystem design

T5.3 Construction of the optimised prototypes

• Manufacturing of the prototypes


T6.1 Performance of test runs with the optimised prototypes

• Performance of a further series of test runs with the adapted and optimised prototypes

T6.2 Evaluation of the demonstration phase

• comprehensive evaluation of the dedicated test runs as well as for the long-term test runs
• The aim of this work is to finally gain a full set of data regarding emissions, thermal efficiencies, emission reduction efficiencies as well as reliability and availabilities of the new technology


T7.1 Techno-economic analyses

• detailed techno-economic analyses for the new technology

T7.2 Comparison of the new technology with present state-of-the-art products

• The data gained from Task 6.1 will be compared with the respective data from present state-of-the-art systems in order to evaluate and quantify the advantages of the new developed system as a whole (including storage, fuel feeding system and combustion system).


Dissemination and exploitation of results

• Creation of a project webpage
• Production of a Video clip, representing the basic concept of the new technology
• interim and final plan for the use and dissemination of knowledge and the exploitation of the technology

Project Results:
Description of the main S&T results/foregrounds

The BioChipFeeding-system can be structured (regarding the material flow) into two subsystems: Feeding Step 1 and Feeding Step 2.
Feeding Step 1 handles the wood chips stored within the wood chip bunker using a gripper system attached to an automatic crane system. Beside the functionality to transport the wood chips from the bunker to Feeding Step 2, the sensors equipped at the gripper provide the ability to analyse the wood chips regarding the properties: moisture content, particle size and color. Besides the screening of this properties, the gripper is used to determine the filling level of the bunker.
Using the wood chip properties and filling level information from the bunker, the implemented algorithm of the feeding control decides according to the current operating status reported by the boiler control which portion of the available wood chips is the most suitable. When possible, the feeding control tries to blend different qualities of the wood chips in order to support a stable combustion of the boiler.
Led by the feeding control, the gripper grabs the wood chips within the bunker and transports them to the hopper of Feeding Step 2, where different qualities of wood chips are blended to a calculated target quality. Continuing through the screw feeder, the wood chips are further transported and dosed by the stoker screw to the firing of the boiler.

Development and Design of Feeding Step 1

Feeding Step 1 needs to cover two main functions:
• Manipulation/transport of the wood chips starting at any point of the wood chip bunker up to the defined material flow interface between Feeding Step 1 and Feeding Step 2 – the hopper.
• Providing information of the wood chips stored within the bunker regarding wood chip quality and quantity.

In order to do so, a system was developed based on the VDI guideline 2221 “Systematic Approach to the Design of Technical Systems and Products” containing two core components: a modular wood chip gripper with built-in sensor technology to provide the needed information and a semi-automatic crane system to move the gripper to its target position within the bunker.

Crane system

Several crane systems from eight manufacturers were evaluated in a methodical decision process leading to the ABUS ZHB system as the most suitable. Though the ABUS crane system offers no automation as a standard, the system was chosen mainly because of its high modularity regarding operable area and load capacity. Other systems, technically on par with ABUS’ system were omitted because of economic aspects.

Table 1 shows a summary of the crane system chosen.

Key facts of the crane system
Mechanical Construction • Aluminium shape
• Fully adoptable to relevant measures
• Modular construction system Agitators • Asynchronous motor
• 400V; 50Hz
• Contactor control provided
• Conductor lines
Crane girder • Length: 4500mm
(up to 12m)
• crane travel: 5/20 m/min
• trolley travel: 5/20 m/min Hoist • Capacity: 250kg
(up to 4t)
• Hoist speed: 2/8 m/min
• Hook path: 3000mm
(up to 32m)
• Power: 0.09/0.35kW

The system was upgraded to provide an automated mode using laser distance sensors and is fully integrated into and controlled by the feeding control.

Wood chip gripper

Based on a detailed specification of the entire heating plant the objectives for the development process as well as the boundary conditions were derived for the wood chip gripper. Using the methodical development process provided by the VDI guideline 2221, starting with the determination of the functions and their structures and carrying on with the search for solution principles and their combinations, a layout of key modules could be developed, leaving enough scope for the installation of the needed sensors. Through its modular design, the developed layout offers the possibility to attach different clamshells to the mainframe and thus providing usability when being operated in systems of different heating capacities. In the course of the project clamshells providing a loading capacity of 35 liters and 75 liters were designed.
Figure 3 in the attached document shows three spotlights of the development process: On the top left side, the identified functions and their structure are shown. The bottom left side of the images shows the found practicable solutions that are able to fulfil the functions. These solutions were evaluated in order to develop the layout of the key modules shown on the right side of the image.
As main actuator for the movement of the clamshells two linear drives built by the company ATP Antriebstechnik were chosen. Each drive provides a force of 6000 newton over a positioning distance of 100mm. The asymmetrical installation position of the linear drives allows the positioning of the moisture content measurement sensor and the image recognition system in the middle of the gripper’s mainframe and thus enables theses sensors to capture the largest possible portion of the woodchips in the gripper, respectively underneath the gripper.
For the operation of the moisture content measurement sensor a second actuator system has been installed in order to provide a constant contact force of 250 newton. The movement of this sensor is performed using a spindle linear table of the manufacturer Igus GmbH. The built-in ball joint allows the system to evenly press the sensor against the loaded wood chips regardless of their topology.

Sensorial concept

In order to provide the feeding control with the needed information a comprehensive sensorial concept was developed, covering the positioning and inclination of the gripper within the bunker as well as the capturing of the wood chip quality. In the list below, the different sensors are assigned to their measurement task within the system:
• Capturing of the loaded wood chip mass:
load cell (Co. ME-Meßsysteme GmbH)
• Determination of the positioning of the gripper alongside of the hoist’s chain:
position underneath the trolley – draw wire sensor (Co. Waycon)
position above wood chip level: ultrasonic sensor (Co. Wenglor)
• Determination of the positioning of the gripper alongside and transversely crane track:
laser distance senor (Co. Wenglor)
• Determination of the inclination of the gripper:
Inclinometer (Co. ME-Meßsysteme GmbH)
• Determination of the moisture content of the loaded wood chips:
moisture content transmitter (Co. Schaller)
• Determination of the particle size and color of the wood chips:
image recognition system (see following section for details)

The two key sensors of the system – the moisture content measurement system and the image recognition system – are described in detail in the following section.

Moisture content measurement

In order to choose a suitable sensor system for the determination of the wood chip’s moisture content, sensors using different physical principles (conductivity, microwaves and time domain reflectometry) were evaluated. Based on a first test series under laboratory conditions a sensor using the conductivity measurement principle was chosen . The sensor is attached to the actuator system described in the previous section. In the course of Task T4.1 (test series without boiler operation ) the sensor was tested under installation conditions.

Figure 6 in the attached document shows an excerpt of the results of the measurement series done for the verification of the sensor. The test series was done using wood samples of different moisture content and particle size with varying contact forces provided by the actuator system. The actual moisture content of the samples was determined using the method specified within the Standard EN 14774-2 represented by the red dots of Figure 6.

Using the developed system, the moisture content of the wood chips can be determined with an average deviation of under five percent points, which is sufficient for the classification of the wood chips as stated in the section Control Algorithm.

Analysis and optimisation of Feeding Step 2

TUM/fml’s task during the project was to analyse and propose improvements for feeding step 2. This feeding step starts with receiving material from the gripper and ends with the transport of wood chips into the furnace. Corresponding boundary conditions for the feeding process were defined in the “book of technical specifications”.

The feeding system can be split into three main parts, which, among others, are shown in Figure 7 (see attachment). They are the hopper (including an agitator/mixer) with a screw feeder at the bottom, a rotary gate valve and the stoker screw feeder, which leads to the furnace. Within the project, the rotary gate valve was not part of the analyses, as it is a standard part and merely used for fire safety issues.
The agitator prevents the wood chips from arching and is mounted above the open trough zone of the screw feeder. As soon as the wood chips leave the area underneath the hopper, the trough consists of a fully enclosed profile. This screw feeder setup was used for experimental measurements of the driving torque, mass flow and effective power consumption. Beforehand, the wood chips had been classified into three different qualities, which were referred to as “low”, “medium” and “high” quality, based on their expected calorific value. For dimensions and operating conditions of the system as well as the measurement results, please refer to the corresponding reports. In order to characterise the wood chip’s flow behaviour, basic bulk material experiments were carried out. These included angle of repose, angle of slip, particle size distribution and bulk density measurements.

All these data, including those from the screw feeder, were used to calibrate and validate a discrete element (method) material model for the “low” quality wood chips. Especially the validation against the results from the hopper/agitator/screw feeder system were computationally demanding. In the end, one simulation of the hopper and agitator system took about 30 hours on a 20 CPU workstation.

Overview of experimental results from the screw feeder

The results from the basic bulk material tests were in good agreement with the literature. It could be verified between the experiments that wood chips, despite lacking cohesion effects, are a poor-flow material which tends to arching. This effect is based on interlocking between the wood chips and abetted by the fibrous nature of wood.

Figure 9 (see attachment) displays different results over time from the screw feeder experiments. These results were measured for the “low” quality wood chips. The hopper was filled with wood chips and agitator and screw feeder (driven by the same motor) were run until the hopper was empty and the screw feeder trough completely emptied (see [Zitat1] for details). In the lowermost part of Figure 9, the effective power consumption of the motor and the mechanical power consumption of the drive shaft are displayed. The latter was computed from the revolution speed and the torque measured at the drive shaft.

Figure 9 (attachment): Mass flow, driving torque and power consumption for the "low" quality wood chips over time, including 95 % confidence interval band for N=8 experimental runs.
Starting from t=300 s in Figure 9, the agitator has lost contact with most of the wood chips, due to the decreasing fill level inside the hopper. After this point in time, the driving torque is dominated by the screw feeder. Therefore, it can be concluded that the torque and thus the power to drive the screw feeder is significantly lower than the power to have the agitator stir through the wood chips. The overall maximum torque to drive both, the screw feeder and the agitator, is roughly 150 Nm, as compared to slightly more than 50 Nm for the screw feeder on its own. Hence, it was decided to investigate on optimised operating parameters for the agitator within the hopper.

Results from discrete element simulations

As mentioned above, the results from the screw feeder experiments and bulk material tests were used to calibrate and validate a discrete element material model for the low-quality wood chips. The discrete element method (DEM) is a numerical simulation method, which can be used to model the flow behaviour of systems of thousands of particles (modelled as spheres or clumps of spheres) and their interaction with bulk material handling equipment; specifically mechanical loads on the latter. Discrete element method primarily demands for CPU time and is usually run in parallel. For large models with hundreds of thousands of particles, the computation duration can be as long as days or weeks.
The material model which was used to simulate the screw feeder and stoker screws is based on spheres. Since wood chips are non-spherical, rolling friction was harnessed to account for the shape of them. This is a common approach in discrete element modelling.

Screw feeder parameter study

As mentioned in the previous section, the overall driving power of feeding step 2 is mainly influenced by the interaction of the agitator with the wood chips. Therefore, a parameter study was carried out. It focused on the positioning of the agitator within the hopper and its aim was to investigate, whether the agitator’s location will affect the driving torque of feeding step 2.
Within the parameter study, the real-life setup of the screw feeder/hopper/agitator system was used as a reference, which is the base to compare different configurations against. The driving torque and mass flow for a total of twelve configurations was computed. The two main factors of the parameter study were the horizontal and vertical position of the agitator inside the hopper and in another configuration the direction of rotation of the agitator was reversed. In order to reduce variance of the results, the computations were repeated at least three times. Please refer to the second BioChipFeeding periodic report or [Zitat4] for a more detailed description of the simulation setup.
Results for the vertical position change of the agitator are displayed in Figure 10. The agitator’s driving torque decreases with an increasing vertical position. This is due to having to stir less wood chips. Even though the risk of arching increases with a higher position of the agitator, no such effect was observed in the discrete element simulations. Simulatively mounting the agitator at a height of 280 mm was the maximum possible translation due to limitations in design and resulted in the lowest torque values. Even just translating the agitator by 40 mm yielded a significantly lower torque demand.

Influence of the vertical position of the agitator on the agitator's driving torque during hopper emptying.
The results from configurations with mutual horizontal and vertical translation of the agitator showed that the positive effect on the torque due to the vertical displacement can be enhanced by horizontally shifting it.

Comparison of two stoker screw design variants

In addition to the parameter study, the validated discrete element material model was used to compare two stoker screw design variants against each other. The stoker screw is the final part of feeding step 2, which feeds wood chips directly into the furnace. It is mounted underneath the rotary gate valve.
Two variants, as depicted in Figure 11(attachment), were compared against each other, with respect to the mass-related energy consumption and possible feeding problems like blockage or jamming for different mass flows. The design on the left-hand side of Figure 11 (Stoker 160) is the new design and features an altered trough geometry and inclination, while the existing design (TBZ 150) on the right-hand side is a horizontal screw conveyor with an upside down U-shape trough.

A comparison between the designs’ respective driving torques is shown in Figure 12. The results for Stoker 160 are displayed both ways, in original and corrected. The correction was necessary, in order to account for the inclination and the resulting lifting work for the wood chips. For the low and medium mass flow (1. and 2.), both variants require about the same torque. However, Stoker 160 demands for roughly 85 % more torque for the high mass flow (3.)

When looking at the mass-related energy consumption, listed in Table 2, one can see that the driving torque is generally lower for the existing design (TBZ150). Hence, the resulting loads in components will be lower, too.
The mass-related energy consumption is lower for the new design. Nonetheless, this only holds true for the lifting-work-corrected values. Given the fact that it is operated with inclination, it loses the energy efficiency over the existing design. Proposed optimisation measures for the new stoker screw design include using it in horizontal orientation and considering to stay with the existing design for high mass flows.
Table 2: Driving torque (median, corrected) and mass-related energy consumption for Stoker 160 and TBZ150; values in brackets show the uncorrected medians.
Mass flow 77 kg/h 93 kg/h 244 kg/h
variant 1. Stoker 160 1. TBZ150 2. Stoker 160 2. TBZ150 3. Stoker 160 3. TBZ150
Torque [Nm] 0.288 (0.773) 0.318 0.484 (0.970) 0.468 1.39 (1.87) 0.746
E’ [mWh/kg] 2.35 (6.39) 4.54 3.95 (7.91) 6.69 11.3 (15.2) 10.6

System Design and Concept of the Overall Control System

Overall Control System

The control system is in charge of controlling the feeding step 1, i.e. transporting woodchips from the unloading area and main storage to the intermediate storage via an automated crane and gripper. The screw conveyor and the boiler use existing control systems and therefore are not part of this description, because there are no further developments to be done in the control systems of these components. As part of the modular approach of the whole concept this can be seen as advantage for the product costs of the system, as it realizes the usage of existing devices where possible. The developed crane and gripper control system is also in charge of picking the correct chips and detecting errors that occur inside the crane and gripper control system. The crane and gripper control system receives information and data from the boiler via an existing Modbus interface. Via this interface the crane and gripper control system receives information about the operational state of the boiler and the actual mass flow. An interface to the image recognition system is used to receive data about particle sizes and foreign materials in the storage, even if the image recognition itself is only included as potential functional add-on in the project.
General Functionality
Once the system is switched on, it is the turn of the worker to choose the mode: automatic, manual or extended mode for troubleshooting. Normally the system will be in automatic mode, but before feeding chips, the initialization step of the system is performed automatically in order to properly taking into account the amount of chips available in the storage and their qualities. Initialization is also obligatory after refillment of the unloading area in the storage from outside and also after other changes inside storage. After the selection of the automatic mode no further user interaction is needed, except the system is shut down or an error occurs. In the automatic mode the crane and gripper system is stopped as soon as an error shows up. The automatic mode reacts autonomously to the boiling process. The actual boiler load is transferred via Modbus to the crane and gripper controlling system. There are several load operation states: ignition, low load, high load and full load plus standby. The control system is setting automatically the priorities to different woodchip qualities and mixtures to be loaded into the intermediate storage regarding these load operation states.
Usually the automatic mode will work like this:
1. Receive information from the boiler about the boiler load state,
2. Analyse available woodchip qualities inside the storage based on pre-sensored information stored in the quality maps
3. Decide which quality or mixture of qualities should be taken,
4. Grip woodchips and transport them to the intermediate storage
5. Update the quality maps every time they change and be aware of errors in the meantime.
Therefore even in automatic mode, a situation can occur that the system is not feeding the intermediate storage, e.g. if the boiler is stopped, working at very low loads or an error occurred inside feeding step 2 or inside the boiler itself. This „spare time“ or stand by is used by the gripper and crane control system to update the quality maps of the storage with additional and refining autonomous measurements and transporting woodchips from the unloading area into the storage. Spare time is detected by filling sensors inside the intermediate storage which are directly giving the information of a full or empty intermediate storage to the crane and gripper control system.
As soon as an error is detected the controlled devices are turned off and automatic mode is finished as part of the prototype safety concept. The error description is shown on the user screen. A qualified worker is needed to access extended operation mode for troubleshooting. This mode is secured via a physical safety key. As soon as the safety key is turned the extended operation mode is activated and the gates to the storage can be passed to get to the crane or gripper system and solve the problem. Before turning the key the whole system is blocked. In extended operation mode all actuators can be directly controlled from the worker, all safety or limit switches are turned off to allow failure debugging. In extended operation mode the control system allows the worker in charge to drive the actuators also to positions, which are not allowed during automatic mode. This means the worker is responsible for system safety and should be conscious of his actions, i.e. enhanced experience with the system is needed. It is also important for the worker to have eye contact to the crane and gripper during this operation mode, therefore the user panel is positioned in a way that the whole storage can be surveyed. Elimination of the failure has to be confirmed by the worker on the user screen. Extended user mode cannot be left before failure is solved, i.e. error no longer detected by the control system and failure elimination is confirmed by user. After this step the safety key is turned back by the worker and user can choose next operation mode.
Additionally a manual operation mode is foreseen, which can be chosen by the user at every time during automatic mode, after system start-up or after successfully finished extended operation mode. In manual mode the crane and the gripper can be moved via direct user inputs, e.g. to drive the crane to a specific position or to grip woodchips at a desired position inside the storage. During manual mode the user is responsible for crane safety, e.g. crashing against objects inside the storage. The manual mode can be used to transport woodchips into the intermediate storage to feed the boiler, but all the information about the quality of the woodchips is not considered for the automatic mode. As soon as leaving the manual mode to automatic mode, which is possible all the time, the automatic mode will “clear” the intermediate storage by letting go down the filling to the lower filling-sensor level and then restart filling it up regarding the actual load state of the boiler.
Hardware Setup
The concept for the controlling hardware consists of one main industrial PC with the central control algorithm and 3 bus couplers with several in- and outputs which are located in 3 different switching cabinets and connected via EtherCAT. For the main controlling unit and the bus couplers as well as for the in- and output cards components from Beckhoff Automation GmbH are used. The following list gives a schematic overview of control units and hardware components for the controlling system. It is divided into three sub controlling units: the main control, the crane control and the gripper control.

Figure 13 (see attachment): basic concept for the overall process control system

The sub controlling units are connected via Ethercat, the main control has interfaces to the user control panel, the image recognition system, which runs on a second industrial PC and the boiler.

Following Sensors and Drives/Actuators were used:

Sensor/Drive/Actuator Task
Light Barrier IELT / IELR Woodchip Level in Intermediate Storage
Laser Distance Sensors DT35 Position of crane in x and y direction
Draw Wire Sensor SX120 Length of Winch / Chain Hoist
Force Sensor KD9363s Weight of Gripper and Load
Ultrasonic Sensor USM603U035 Distance from Gripper to Woodchip Surface in storage
Tilt Sensor ME IS 28 Inclination Angle of Gripper (x and y direction)
Linear Actuator LMR03 Closing / Opening the Gripper
Moisture Meter / Humimeter Moisture of Woodchips
Stepper Motor NEMA23XL Pressing Humimeter onto Woodchips inside of the Gripper

A detailed overview of all the components is given in Deliverable D2.3 – Detailed design of the process control.
Safety Concept
For safety issues the controlling system is equipped with certain functionalities to prevent personal injury and property damage. Therefore a safety controller and safety in- and outputs are integrated into the controlling system. If a person is entering the storage the crane system and the gripper are disconnected from the power supply and the winch for the gripper is mechanically blocked. A safety light barrier is integrated into the entrance gate of the storage to detect safety relevant situations and shut off the crane. In addition several Emergency Stop Buttons are installed at the user control and at the access points to the storage to shut off the system in emergency situations.

Circuit Diagram

The configuration of power supply splits in several parts, including the main control cabinet for the industrial PC, the control cabinet on the Abus crane and the control cabinet on the gripper. The power supply for the control system is separated from the power supply for the crane and the gripper to prevent any interference. To avoid current peaks in our control circuit, which can occur when the linear motors or the stepper motor on the gripper are switched on the power supplies of the gripper motors and of the controlling components are electrically separated from each other. The detailed description of the power supply concept can be found in Deliverable 2.3.

Control Algorithm

The control software is implemented using Twincat V3, an IDE and runtime provided by Beckhoff Automation and widespread in the industry. It uses IEC 61131-3 standards and all the controlling code was written using an object oriented approach to ensure simple adjustability for the later use by the SME project partners.
The process starts with an initialization of the system. After that the user can choose between the manual and the autonomous, i.e. continuous, operation mode. The autonomous operation mode as major outcome of this project is described in the following paragraph, for further details on the overall control strategy see also Deliverable D2.3
In autonomous operation the algorithm handles the monitoring of the storage, the material selection and transport, updating of the quality maps and the intelligent feeding of the intermediate storage. First the control unit communicates with the boiler to obtain the current operation (ignition, stand-by or one of the three continuous modes: low, high and full load). During the ordinary operation the storage is fed with biochips, if the current mass in the intermediate storage falls below a predefined level.
During the continuous operation it is necessary to derive the area that is tangible for the gripper due to restrictions of the topology of the storage. Next the tangible area is further restricted based on the quality of the wood chips. The algorithm computes the suitable qualities, defined by their ash- and moisture content. The idea is to limit the qualities to those, that can be fed to the intermediate storage and keep its parameters within certain limits. Finally the allowed qualities are superimposed with the height profile of the previous step. The so obtained tangible area lays the basis for the steps to come. This includes the feeding of the intermediate storage with bad qualities, with material featuring the needed priority or finally by mixing the available materials.
The algorithm tries to get the material with the highest priority value, depending on the loadmode of the boiler. In this case priority ranges from 1 to 15 with 1 as the highest priority. Initially it checks if material defined by the recent priority which results in the desired quality after mixing in the intermediate storage is present. If so the algorithm primary uses bad quality material (e.g. high moisture and high ash content). If no suitable material is present the algorithm tries to mix the material with the desired priority value from two components. It first feeds the worse material, the better later on. Due to the results of the evaluation of the prototypes some adaptions to the underlying priority maps for the algorithms where defined and implemented to the control algorithm (Figure 14 in the attached document).

Details can be seen in Deliverable D 6.1 results of tests with the optimized system.
After the selection of the material quality to grab the handling strategy for storage management decides where to grab the material in the storage in case a specific quality is available in serveral locations inside the storage. This selection process is realized respecting several conditions: flatten the storage topology to increase potential positions to grab material and reduce ways for navigation in storage. In spare time the picking routine is also be used to transfer material from unloading area to main storage. For the deposition of material the process for emptying the storage is basically reversed. The goal is to continuously fill the storage starting from a defined position. If possible material of the same quality is placed nearby. Moreover areas whit too large gradients are prevented.
Detailed information about the algorithm, implemented class diagram and underlying risk analysis can be found in Deliverable 2.3.
Data recording and Visualization is directly integrated in the graphical user interface of the system. System data can be visualized directly on the integrated 19” touchscreen of the system or be exported in standard formats, e.g. Microsoft Excel. Recent Activity of the system is presented to the user on this touchscreen in form of a graphical animation of the crane and in textual form. The user interacts with the system via software buttons on this touchscreen.

System Design and Algorithm Description of the Image Recognition System
Objectives and Tasks of Image Recognition System
The main purpose of the Image Recognition System is to provide the control system with information about the visual properties of fuel at a certain location within the fuel storage. These properties have been specified to be the average wood chip particle size at the surface as well as the mean fuel quality at this location. Further tasks are the detection of oversized chips or foreign objects which might damage the stoker or obstruct the grate, as well as the detection of large accumulations of saw dust. In these cases the Image Recognition System alerts the control system of its findings.
Sensor Architecture
The Image Recognition System consists of two GigE Vision industrial Ethernet cameras, two 100W LEDs as light sources driven by an Ethernet capable flash controller and a computer for image acquisition and evaluation. Auxiliary components are an Ethernet switch connecting the cameras and the flash controller to the image acquisition computer and two DC-DC switching power supplies providing the components with their required voltage from a common 24V power line.
The camera system operates independent from the control system. After power up, it enters an idle state waiting for a trigger signal from the control system, upon which it acquires a sequence of images illuminated alternatingly by LED1 and LED2 and generates an exposure sequence. After evaluating these images statistics are transferred back to the control system. All operations are performed autonomously, the only interaction necessary between control system and image recognition system are the initiation of the image acquisition (control system → image recognition system) and the subsequent feedback of the evaluation metrics (image recognition system → control system).
Algorithm Description – Segmentation
Woodchip particle segmentation is performed with an adapted shape-from-shadow approach motivated by the shape detection algorithm proposed by Raskar et. al. for their non-photorealistic camera. To ease segmentation we strive to acquire images in which the chip area is overexposed up to the exception of the cast shadows arising from lateral illuminating the scene. The resulting image pair {ILED1, ILED2} is fused into a single image IMIN by taking the pixel wise minimum of {ILED1, ILED2}. Individual wood chips are segmented by seeded region growing the distance transformation of a binarized IMIN. The results are refined by separating the segments into star-convex blobs.
Segmentation was tested on image sequences acquired both in the lab as well as during the test runs at the BIOS prototype plant. Since no ground-truth data is available for the segmentation task only qualitative evaluation was performed.
Algorithm Description – Sawdust Detection
Sawdust detection is based on the statistics of the segmentation without post processing. The sawdust flag is raised if the segmentation results in a few connected components where the bounding box covers a large area of the image, yet the filled areas of these connected components are considerably larger than the component areas themselves. Small areas with a size comparable to that of individual wood chips are ignored for sawdust detection.

Figure 15 Exemplary result of the sawdust detection. Left: segmentation result without star-convex separation applied motivating the criteria for classification as sawdust. Right: Illustration of the detection.
Algorithm Description – Woodchip Size Estimation
Metric information is obtained by fitting a plane into a subset of the 100 best matching 3D point correspondences from sparse stereo reconstruction of the fuel surface. The segmentation is then projected onto that plane and by this pixel areas measured in pixels are transformed into a metric measurement in mm². Since the control system utilizes the mean chip size only and due to the flat-shaped nature of wood chips this approximation is sufficient.
Algorithm Description – Overlarge Wood Chip Detection
An overlarge object is detected and the corresponding flag raised if the segmentation yields an individual chip with an area or a major axis length exceeding a predefined threshold. This requires the objects surface and texture to behave in such a way that it appears as a single contiguous area in the segmentation. Overlarge particles of the same substance as the fuel (i.e. wood in most cases) can be expected to meet this constraint. Particles of foreign material will be detected by the foreign object detection. Since handling is the same for both overlarge wood chips and foreign material a detection by both algorithms does not pose a limitation to the performance.
Algorithm Description – Quality Estimation
The fuel quality is tied to the optical fuel brightness. Bright fuel is thought to contain just a small amount of ashes and be of high quality, dark fuel with a large relative bark content is thought to be of low quality with a rather high ash content. In order to assess the fuel brightness reliably the radiometric response functions of the sensors are calibrated according to the method proposed by Debevec . Fuel quality is then determined by calculating the radiance map of a fuel sample and classifying it according to the radiance histograms of images of fuel samples of the predefined quality classes.

Tests have shown that comparison of the median radiance is enough to classify fuel samples correctly. As one would expect, a wet sample from a certain class results in a lower radiance measurement. In most cases however correct classification was still possible, especially so when restricting radiance measurements to segmented areas only.
The algorithm was tested during the scheduled test runs at the BIOS prototype plant.
Algorithm Description – Foreign Object Detection
The objective of the foreign matter detection is to alert the control system of the presence of foreign objects on the surface of the fuel pile with a size that may either jam or even damage the stoker screw or obstruct the grate of the furnace to an extent that may impair the performance of the boiler. It is not necessary to detect every foreign particle and obviously it is not possible to detect foreign matter completely covered below the fuel surface.
Foreign object detection is based on sparse coding of image patches, following the algorithm proposed by Lu . A foreign object is detected and the corresponding flag raised if a certain image area cannot be explained by sparse combinations of pre-learned bases vectors with just a reasonably small reconstruction cost.
The sparse bases combinations were trained on images of various wood chip sizes and qualities, testing was performed on images of the same wood chip samples with metal objects (different bolts, a ring cut from a circular piece of pipe, and a steel strut) placed in them. The foreign objects were in some cases partially covered by fuel, in other cases they were completely visible.
A further limitation of the current system is its inability to correctly handle previously unseen types of biomass fuel. This may be tackled by providing the user with an option to start a re-training phase or alternatively by an online updating scheme.

Figure 19 Sample images of the test objects and some of the fuel types used for training respectively testing the foreign object detection.

Potential Impact:
Impact of the new technology

The advantages of the biochip feeding system compared with the walking floor system, which represents the state of the art, are from financial and technical nature. Both aspects are from high importance for the enduser. The new feedingsystem opens the possibility for the controle of the boiler to order the most suitable material for the momentanous state of operating, dry and nearly ashfree material for full load, for ignitition and for minimum power, fuel with high water contents, for high load from 60 to 85 %. Fuel with high ash contents should be burned in low power operating states to avoid the melting of the ashes, what means slagging. If the ordered material is not available, the system is able to mix materials most similar to the ordered quality. Worse material should be used as often as possible, the quality of the fuel in the storage should increase. The biochip feeding system is based on a crane with a gripper, which takes the fuel, from the top of the woodchip piles in the storage. In reason of the low mass of fuel used per hour – this is about 150 kg for a 400 KW boiler- the gripper and the crane could be a lightweight construction, which means a low electrical power effort for the drive motors. The whole conveying technology is always above the material, in case of a reparation it is easily reachable for service technitians. The fuel need not to be removed like in plants representing the state of the art. Also the planning of plants in non-even areas is much easier, height differences are not so problematic as for the planning of a walking floor system. The distance sensor allows a signal to the end-user to fill up, when the storage gets nearly empty.

HDG Bavaria will create with the results of the biochipfeeding project a new product. A new and intelligent conveying-system with many unique selling points. These unique features will create additional turnover by selling the new system, but also by selling boilers, which would be not sold if only an old fashioned conveying system could be used. Therefore the biochipfeeding system is a key for selling more boilers. The sophisticated controle system gives HDG, but also the customer of HET GmbH the image of technical leaders, a so called hitech image, but also a positive environmental image caused by lower emissions, higher efficiency and lower demand of electricity. The sale of the biochipfeeding will redeploy the endusers investment for concreting (the floor of the storage) to business volume for HDG by selling the gripper system. Against the heavy weight design of state of the art conveyors, the biochipfeeding system seems to be a lightweight construction, which means lower transport costs, lower manipulation costs less manpower for installation. The position of HDG but also of the future customer of HET GmbH against the competitors will rise.

For HET the results of the biochipfeeding project are a possibility to sell a developing project - a seriaproduct - developing -project to one of the great woodchipboiler producers- one, because HET and HDG made an agreement that HET will feed only one of HDG`s competitors with this new technology. Another possibility of the biochipfeeding technology is the earning of royalties.
Generally new technologies are important for the image building of HET GmbH. The image of creating high grade technology clears the way to new customers – so the biochipfeeding technology is a sort of marketing instrument.

Sinte srl will use the results for creating and producing a boiler controle plate including the demands of the biochipfeeding system. The costs for a woodchipboiler controle system are about 1000,- to 1500,- € ,acquisition price for the boiler producer. A new plate for the controller of boiler and biochipfeedingsystem will have a price about 1300, - to 1800, - €. This opens Sinte the possibility to create new turnover of 70.000 - to 100.000,-€ per year and the chance to find new customers, who –maybe- will order also another merchandise wares.

The main benefits for endusers are a cheaper investment for the complete plant- especially for concreting of the floor of the fuel storage. But also, as mentioned above, the lower costs for service – the conveying technology is above the fuel piles-, lower costs for the operating and fuels. For instance you can mix agricultural fuels with high ash contents like straw consequently but in small quantities to fuels of high quality. In some cases an installation of a woodchipboiler plant could be only possible with the constructional flexibility of the gripper solution. For local heat providers the possibility of fast ignition and a fast uprating of the power after start by using dry fuel is one of the most important assets, they cannot accept long delays for getting warmth into the district heating grid. The possibility to recognize and remove foreign or oversized parts will reduce the number of shutdowns of the plant, although there is no 100% probability to find all of them.

impact to the market: the best is the enemy of good, so there is no doubt that the biochipfeeding will be a success and a new leveling for the competitors. The presentation of a fully developed seriaproduct will be at the ISH 2017 in Frankfurt, the most popular fair for heating systems in Europe. We think in 2016 will be a market recovery for woodchipboilers in 2016. The law “Bundesimmisionschutz - BIMSCH” in Germany and the reduced fundings in Great Britain slowed down the sold quantities of woodchipboilers in 2015. So the commercial introduction in 2017 will be at the best moment. The new system will be introduced from two producers, one will be HDG Bavaria, the other one the customer of HET GmbH – in the moment there are negotiations with a wellknown Austrian boilerproducer. If more producers are selling the same new and innovative kind of technology it avoids an exotic image and increases the credibility of this new system. And more providers with the same technology represent more power on the market for a new idea. HET and HDG think about two versions of the biochipfeeding system, the premium version using moisture measurement, image recognition, distance sensor and a standard version without image recognition, only respecting the water contents of the fuel. What is the sense of such a reduced version? A lot of endusers of woodchipboilers combine a boiler in the powerrange about 400 KW with silotechnology without sufficient robustness,like agitators with articulated arms. The reason is the small amount of investment, compared to a walking floor system. A 400 KW plant needs about 1500 to 2000 m3 woodchips per year, the stirring units are dimensioned for 400 m3 per year for a livetime of 20 years. After four or five years of intensive use of stirring units, these systems are completely at the end of duration. If it is possible to position a standard version of the biochipfeeding system in price in the middle of the gap between walking floor system and agitator, many endusers will tend to the gripper system. That means in average more turnover per sold plant.

Law and fundings: The Energy Labeling Directive, presented in 2015, needs solutions like the grippersystem, the lower amount of electricity allows one or two steps upwards. From other technological sectors like refrigerators, television or washing machines the energy labeling is well instituted in the minds of the citiziens of the European Union. The upgrade in the ranking will sometimes be the last determining factor for the purchase decision.
The lower emissions of the boiler and the higher efficiency could be the last step to undershoot the standards of the different national funding regulations. After the last use it is easy to disassembly the crane and gripper, also the recycling process seems to be a simple challenge. In a lot of countries the standards for dustemissions became stricter. To maintain these standards secondary measures like static dustfilters are the last possibility. These dustfilters are effective but expensive, with the optimization of the combustion the use of filters could get needless.

Public economic impacts: Woodchipboilers- although from a heavy and robust look- are not “Low Tec” products. Most of the producers weld and mount their products in the European Union, what implies a high part of valuecreation in the Union, often in the home countries. The sale of woodchipboilers inclusive feeding- and silotechnology ensures jobs in different manners.
The technology is produced in the Union, the installer comes from the region, the fuel comes from the region, sometimes of the own wood of the enduser. All connected taxes are paid in the Union.

environmental impacts: As we could feel in the last summers, the climate change briskly accelerates – the CO2 emissions should be drastic reduced, but harmonized with a sustainable forest management, which guarantees the often mentioned CO2 neutrality. The improvement of the combustion causes lower emissions from a incomplete oxidation, as the CO-emissions, hydrocarbons and the dust. The higher efficiency decreases the absolute CO2 emissions, because less fuel has to be burned to get the same output.
Another point of view - woodchip boilers, woodchips storages and conveying systems are never involved in incidents compareable to oilpolution or gasleakiges.
The gripper and the crane are light weights against the waking floor systems, a mentionable saving of raw material can be claimed for the biochipfeeding system.

Plan of dissemination:

HDG: The crane system will be a high technology product. Not every HDG sales man will be able to sell it. HDG installed special project managers in Germany. They sell special products like HDG M300-400 and in future the crane systems. HDG has already implemented these new salesmans. After finishing the development of the seriaproduct and presenting it at the fair in Frankfurt in 2017 these salesman can offer this new product.
In a best case – worst case consideration, HDG will increase the turnover from 5% to 10 %, which means 1,5 Mio € to 3,5 Mio € per year.
The costs for the mentioned fair in Frankfurt the ISH will be about 60.000,-€ , the costs for the smaller but very important fair „Energiesparmesse“ in Wels/ Austria will be about 40.000,- €.

HDG will implement the optimisations of the auger design done by TU-München fort he feeding step 2 in all current HDG Products, also in the normal actuatorsystems and conveyingsystems. This process has already started.

The question of a patent application is not decided yet and will be discussed shortly between the SME`s, a advance in time will be sometimes more helpful than patenting.

In several journals (as HLK – Heizung, Lüftung, Klima ) and newspapers (Salzburger Nachrichten, OÖ-Nachrichten) have been summaries and short presentations of the biochipfeedingproject, HET got some reactions of potential partners. A presentation of TU Graz was held at the end of octobre 2017 in Styria, at a symposium of local warmth providers.

What are the intentions of HET
HET wants to develop the final version of the grippersystem for a woodchipboiler producer
HET wants to create a turnaround from this project about 350.000,- to 500.000,- €
HET will do the technical design and all drawings for the industrial version of the biochipfeeding-technology
HET will complement and rework the software for the controle to an industrial version
HET wants to do the necessary tests to reach marketability
HET wants to do the certification for the admission for the market
HET wants to creat a turnover from the licenses in the height of 100.000,- € , 20 plants per year about more than five years, 1000,-€ per plant

The agreements between the SME`s have to be revised together and changes should be done, if some points are obsolet or some facts have changed.

List of Websites:
The project’s website is available using the link
The website is structured into seven pages:
• BioChipFeeding (Start page/landing page)
• The Project (Information about the project and its main objectives)
• Consortium (List of all consortium members, including contact information)
• Project Meetings (Information about the project meetings, stated in the list of deliverables – all other meetings are not listed)
• Concept Illustrations (Videos and Images, illustrating the basic concept of the project)
• Prototype Facility (Information about the prototype plant in Massing, Germany)
• Conferences