Skip to main content

On-line early damage diagnosis, prognosis and root cause analysis for Industrial multi-stage gearboxes used in the water industry

Final Report Summary - INDGEAR (On-line early damage diagnosis, prognosis and root cause analysis for Industrial multi-stage gearboxes used in the water industry)

Executive Summary:
Project IndGEAR: On-line early damage diagnosis, prognosis and root cause analysis for Industrial multi-stage gearboxes used in the water industry.

The balance between water demand and availability has reached a critical level in many areas of Europe, according to the European Environment Agency and it is expected that the situation will be exacerbated in the future by climate and social change. The UN Environment Programme reports that trends indicate a 40% rise in average water consumption by European residents over the next 20 years, while EC statistics show that total freshwater resources are at relatively low levels (below 3,000m3 per capita) in the six largest Member States (Germany, Spain, France, Italy, Poland and the UK), as well as in Belgium, Denmark and the Czech Republic.

It is clear that the treatment of wastewater has a key and growing role to play in addressing this problem and will require an increase in the number of water treatment facilities available. European wastewater treatment plants delivering high-quality clean water from sewage face a number of unresolved problems related to gearbox operation. Because most gearboxes operate in outdoor conditions, they are one of the most vulnerable units at treatment plants. The following conditions can lead to extensive gearbox damage/failure:

* The probabilistic nature of gearbox loads and essential load variation during gearbox operation.
* High levels of humidity increases the possibility of water ingress into lubricant oil of the gearbox. • High temperature variations.
* Most gearboxes operate in non-stationary start/stop conditions.

The lndGEAR project — through our consortium of SMEs, research and development partners, and major water industry end users - aims to introduce the radically novel triple technology for damage diagnosis, damage prognosis and root cause analysis through radically novel signal processing of gear vibrations.

Within the scientific objectives, we have successfully developed novel models and dependencies between time-averaged gear vibrations and stresses in gearbox wheels, as well as novel generalisation of the Palmgren-Miner rule for improved accuracy of gear damage prognosis.
The consortium have realised, within technical objectives, novel non-stationary digital signal processing techniques for adaptive detection of non-stationary impacts due to damage, and early differential automatic damage diagnosis, prognosis and root cause analysis technologies for gearboxes. All technologies for on-line diagnosis, prognosis and root cause analysis have been successfully validated using laboratory test rigs and in-field experiments with gearboxes, at waste water plant. In addition, the developed hardware platform and communications methodology/system is based on open system architecture platform enabling integration of damage diagnosis, prognosis and root-cause analysis technologies.

THE PROJECT WEB PORTAL is available to the Project Consortium and the Home Page is also available to the General Public at: http://www.indgear.eu
Project Context and Objectives:
The IndGEAR project − through our group of SMEs, research and development partners, major water industry end user and global service provider to European water industry− aims to introduce the radically novel triple technology for damage diagnosis, damage prognosis and root cause analysis of damage through radically novel signal processing of gear vibrations.

In simple terms, this means introducing radically new approaches/techniques to processing vibration signals collected from gearboxes while they are in service.
In technical terms, this means using the novel non-linear non-stationary higher order spectra and spectral correlations and novel non-stationary spectral kurtosis to advance the state of the art in signal processing/vibration condition monitoring.

These novel condition monitoring technologies will:
* Improve the utilisation of gearboxes and prevent unnecessary downtime of machinery, thus greatly increasing plant efficiency and reducing life-time costs.
* Provide understanding of root causes of gear faults/failures, enabling proactive maintenance and design to be based on removing those root causes, and, preventing future recurrences of faults/failures.
* Provide estimations of the remaining useful life of gears, enabling better planning of maintenance actions and, therefore, avoiding catastrophic gear failures due to fast damage propagation.

PROJECT OBJECTIVES:
Scientific:
* Develop via Simulink modelling and finite element analysis and experiments, novel models and dependencies between time-averaged gear vibrations and stresses in gearbox wheels.
* Develop novel generalisation of the Palmgren-Miner rule to improve accuracy of gear damage prognosis.
* Develop novel non- stationary digital signal processing techniques for adaptive detection of non- stationary impacts due to damage, diagnosis of non- linearity in impacts and gear resonances and online technology adaptation. Research into anomaly detection techniques for early gearbox diagnosis
* Validate novel approaches for on-line diagnosis, prognosis and root cause analysis via controlled test rig and in-field experiments with gearboxes.

Technical:
* Develop novel in-service adaptive early differential automatic damage diagnosis technology for gearboxes.
* Develop novel in-service early differential automatic damage prognosis technology for gearboxes.
* Develop novel in-service differential automatic root cause analysis technology for damage in gearboxes.
* Develop open system architecture software platform to integrate damage diagnosis, prognosis and root-cause analysis technologies.
* Develop hardware platform and communications methodology/system.

Integrated:
* Integrate open system architecture vibration damage diagnosis, prognosis and root cause analysis system for gearboxes.
* Validate (demonstrate) the developed system via test-rig tests and in-field experiments at waste water plant.

DESCRTIPTION of INDGEAR TECHNOLOGIES
Novel signal processing technologies, which include An automated Time Synchronous Average (TSA) Technique, Improved gear residual signal estimation methodology, Optimized Spectral Kurtosis (SK) technique, Wavelet Spectral Kurtosis technique are implemented and validated for fault diagnosis of single stage and multistage industrial gearboxes. In addition, Advanced decision making technologies are implemented to avoid false alarms in final diagnosis decision making, namely Anomaly detection (Stage 1), Damage Detection (Stage 2), and Diagnosis decision (Stage 3).

Novel signal processing techniques are implemented for the root cause analysis of gear tooth damage. The algorithm of root cause analysis technology is developed. The integration of MATLAB codes of developed Cranfield’s technologies into Lab-View software is performed for convenient interpretation of results for the end user.

The prognosis technology implemented aims to estimate the remaining useful life of the gearbox (RUL) by using a damage accumulation law. The damage accumulation law uses the material S-N curve and the number of fatigue cycles accumulated during the different resonance crossing, detected via Chirp-Fourier Transform. It is generally required a previous study on the resonant frequency by using a finite element model to determine the frequency-rotational speed dependency law.
Project Results:
GENERAL SUMMARY
Cranfield presented a comprehensive state of the art in the signal processing techniques suitable for the non-stationary impacts and gear. The consortium has completed all the research related activities pertaining to the development of IndGear Condition Monitoring system. Cranfield implemented novel techniques to perform diagnosis; prognosis and root cause analysis (Imbalance and Misalignment detection). Gearbox fault diagnosis has been successfully performed by implementing Cranfield’s novel signal processing technologies. The schematic of the technology is developed and algorithms of the main steps of the technology are also developed. The improved spectral kurtosis method is implemented, based on gear fault feature extraction performs effectively; this approach is validated using simulated and experimental data of gearboxes.

Novel signal processing technologies, which include An automated Time Synchronous Average (TSA) Technique, Improved gear residual signal estimation methodology, Optimized Spectral Kurtosis (SK) technique, Wavelet Spectral Kurtosis technique are implemented and validated for fault diagnosis of single stage and multistage industrial gearboxes. In addition, Advanced decision making technologies are implemented to avoid false alarms in final diagnosis decision making, namely Anomaly detection (Stage 1), Damage Detection (Stage 2), and Diagnosis decision (Stage 3).

Novel signal processing techniques are implemented for the root cause analysis of gear tooth damage. The algorithm of root cause analysis technology is developed. The integration of MATLAB codes of developed Cranfield’s technologies into Lab-View software is performed for convenient interpretation of results for the end user.

The implemented prognosis technology aims to estimate the remaining useful life of the gearbox (RUL) by using a damage accumulation law. The damage accumulation law uses the material S-N curve and the number of fatigue cycles accumulated during the different resonance crossing, detected via Chirp-Fourier Transform. It is generally required a previous study on the resonant frequency by using a finite element model to determine the frequency-rotational speed dependency law.

Data acquired in laboratory conditions and infield conditions are processed using the proposed techniques for performing diagnosis, prognosis and root cause analysis. Under laboratory conditions experiments that are performed at University of Modena (Test Rig 1 and 2, Simulation data), and ERIK’s Test facility, the data are processed and technologies are successfully validated. All the tests are performed on single or two stage gearboxes and accelerometers are used to capture the vibration data.

The MATLAB programs developed by Cranfield are integrated to LabView Software environment. The IndGear hardware and software successfully been integrated into the prototype: i.e. panels built for field testing. The LabVIEW user interface software has also been integrated and deployed on the Human Machine Interface (HMI) touch panel computer.

During the second year the group of Modena was mainly focused in the following activities: modelling development for root cause analysis purposes; experimental activities for validation and database production; dissemination activities. Phenomenological models were developed and validated, these models furnish both quantitative and qualitative analyses: stress evaluation, damage evolution (from experiments), intrinsic properties of the gear vibrations and their interaction with the surrounding mechanical elements (shafts, housing, joints, motors).

The methodologies have been developed for an industrial usage, taking advantage from the newest analysis approaches available in the scientific literature, which have been converted to the specific needs of a engineering CM analyst. All activities have been object of a reporting having also the goal of furnishing the technical background for the use of the methods.

An extensive experimental activity was carried out for setting up models, collecting data for validation purposes, extracting system properties (this includes the development of suitable experimental methodologies).

The main activities in the second year of the project for LUKA was related to analysis of test rig test results conducted by Modena University. Result of test runs gets to conclusions for further steps. We have agreed with Modena to prepare new design of the gear pair for test rig 2 in Modena’s lab in relation to pitting occurrence at 800rpm and approximate torque 140Nm. The new MathCad calculation and design were made in February 2016 and send to Modena for their FEA comparison.
The main results related to technologies development include:
1. A novel technique to estimate the instantaneous frequency of rotation from the acquired vibration data in absence of speed sensor is implemented.
2. For performing imbalance detection for gearbox vibration signals that are non-linear and non-stationary, a novel signal processing technique, short-time higher-order chirp transform (STHOCFT) is implemented.
3. Bicoherence technique is implemented by integrating with STHOCFT for misalignment analysis for

diagnosis using non-linear and non-stationary gear signals.
4. A novel anomaly detection technique is implemented based on kNN classification, method, k-means

clustering, novelty scores, and weighted majority rule.
5. Root-cause analysis technology based on novel signal processing techniques is

implemented on gear vibration data for all shafts of the gearboxes.
6. Software for diagnosis, prognosis and root-cause analysis technology is prepared.
7. Software Specification – is created
8. The technologies that have been developed in MATLAB are transferred to DLL file formats which are integrated with the LabVIEW user interface software developed. The results produced from the DLL files are made accessible to the user interface software in a defined format that the software will use to interpret the results.

DETAILS on WP TASKS and ACTIVITIES

WP1
Cranfield collaborated with MODENA to agree on test parameters such as input speed and load to perform the experiments. Also commented on the FEA simulation model to make it more realistic. Northumbrian water commented and advised on the experiments performed. ERIKs suggested test parameters such as input speed and load to generate excessive vibrations. These parameters are used to design experiments suitable to validate Root Cause Analysis experiments. ISRI involved in decision making related to the data acquisition system and the data acquisition software. The group of Modena completed their activities of this task, which was concluded at month 14. Two test rigs were used for this activity both for stress analysis and endurance tests: the test rigs were instrumented with accelerometers, torquemeters, Laser Doppler and strain gauges. Endurance tests were monitored in order to produce high quality data to be processed with CM algorithms: experiments were successful as they produced perfect pitting patterns and reliable vibrational data. The SME have been present during this task mainly in terms of consulting and advisory activities.



For this task Cranfield processed the experimental vibration data obtained from MODENA gear test rig. For diagnosis of the gear tooth pitting faults Cranfield proposed to implement Classical and Wavelet Spectral kurtosis techniques. A comparative study between both the techniques has been undertaken. Also novel decision making technology based on weighted majority rule is validated. The total probability of correct diagnosis is proposed to be estimated using the proposed decision making technique. The total probability of correct diagnosis was estimated to be 100%. The validation of the Root Cause Analysis (RCA) technology for the imbalance and misalignment detection was demonstrated using the data from the Modena gear test rig. The imbalance detection results were obtained using the magnitude of the short time chirp Fourier transform at the first order frequency of shafts 1, 2 and 3. The misalignment detection results were obtained for the Modena test rig data using magnitudes of the classical bicoherence components 1-2 and 1-3 for shafts 1, 2 and 3 of the gearbox. The total probability of correct diagnosis was estimated to be 100%. During task execution, Cranfield has received valuable support and revision of the activities from the all the SME partners.



For online adaption of the proposed techniques, Cranfield implemented three techniques during various stages of technology development. The novel online adaptation techniques developed as listed as follows,
1. Adaptive CSK technique.
2. Adaptive Constant Resolution Wavelet Spectral Kurtosis (CRWSK)
3. Adaptive Short Time Higher Order Chirp Fourier Transform (STHOCFT).
For gear damage diagnosis, the adaptive classical Spectral Kurtosis (CSK) technique is implemented. A thresholding technique used for diagnosis of gearbox tooth faults is introduced at this stage. The diagnosis procedure involves,estimation of the Time Synchronous Average signal and the gear residual signal, and then the Spectral Kurtosis adaptive optimal filter is estimated using the thresholding procedure.
By considering overlapping among the TSA segments, several realizations of the TSA signal are estimated. It is very important that the SK estimated over the realizations should be consistent. The statistical SK thresholding procedure presented in literature is used for comparing the proposed approach. A three stage diagnosis decision making technique based on weighted majority rule is used for final diagnosis.
Along with classical Short Time Fourier Transform based SK technique (CSK), adaptive Wavelet Spectral Kurtosis (WSK) technique is also tested to compare the performance of the proposed technique. The constant resolution wavelet SK ensures that the optimal filter is estimated, exactly corresponding to the signal transient impulses, in other words the CRWSK adapts to the resolution of the signal under inspection.
Variation in the speed cannot be avoided for a gearbox subjected to heavy duty operation. The higher order Chirp-Fourier spectra (HOCS) was proposed and investigated by Cranfield for online adaption to the variation of the instantaneous excitation frequency. The proposed adaptive techniques are integrated into the RCA technology for imbalance and misalignment detection. .
The STHOCFT transform is suitable for the processing of the gearbox vibrations at non-stationary regimes of operation, i.e. with variable speed.
The proposed multidimensional HOCS is suitable for the above mentioned signals because
* The non-traditional transient kernel of the higher order Chirp-Fourier transform is suitable for transient signals with any polynomial frequency variation.
* The time-domain windowing technique is suitable for transient nature of signals.
During task execution, Cranfield has received valuable support and revision of the activities from all the SME partners.



RCA technology is implemented for imbalance and misalignment detection. Novel second and higher order spectral techniques such as Short Time Higher Order Chirp Fourier Transform and CFT Bicoherence are successfully implemented using the magnitudes of these transforms.. However, in order to further improve probabilities of correct diagnosis, the new approach of the amplitude-phase extraction from these techniques was implemented and successfully experimentally validated. This approach improved the effectiveness of diagnosis comparing with the magnitude approach. During task execution, Cranfield has received valuable support and revision of the activities from all the SME partners.



The implemented anomaly detection method (Cranfield Background IP) is the supervised classification method based on the modified non-parametric kNN approach. The implemented method consists of the following steps:
* data clusterization - training,
* calculation of the novelty scores – testing,
* decision making for single anomaly detection.
For training purposes, the data were prepared by extracting the features from the undamaged case. This is followed by establishment of the training clusters using the k-means method, and calculation of the averaged distances to k nearest neighbours (the averaged kNN distance) for each sample in each cluster of the training data. Decision-making procedure is based on the comparison of novelty scores with a detection threshold. The test data sample is believed to be the single anomaly when all novelty scores exceed the distance threshold, otherwise it is believed that no anomaly detected. The proposed anomaly detection approach is able to perform decision making independently for each tooth, as the diagnostic features are extracted for each tooth as well.
In addition to anomalydetection, a three stage decision making based on weighted majority rule is implemented. This technique is designed to identify and to remove false alarms generated during the anomaly detection stage.
The vibration data obtained from MODENA test rigs were processed using the decision making technology. For damage diagnosis using spectral kurtosis and for root cause analysis via CFT and HOS features the decision making technology was observed as highly reliable providing an essential improvement of the total probability of correct diagnosis. During task execution, Cranfield has received valuable support and revision of the activities from all the SME partners.



Nonlinear FEA models were developed in order to evaluate strain and stresses on gears, these models were combined with new semi-analytical models in order to take into account the dynamic effects. A combined experimental and modelling procedure is developed to obtain the stress vibration dependencies: experimental vibration data is used for setting up the dynamic model (in particular the dissipation model), the dynamic model furnishes to FEA the actual loading conditions, the FEA evaluates stresses and contact pressure on the gears. Northumbrian water commented and advised on the finite element modelling and reviewed the final FEA model.
Cranfield advised and reviewed on parameters related to the FEA model developed to obtain the stress vibration dependencies. Cranfield also involved in the design of experiments related to this task.
The SME partners have provided valuable support and revision of the activities.

WP2
Cranfield implemented the technology for diagnosis of gear faults. Along with Classical Spectral Kurtosis, Wavelet Spectral Kurtosis (WSK) with constant resolution has been developed. It is observed that the diagnosis effectiveness obtained using WSK is higher than for using the CSK.
Feature extraction is proposed to be performed in two main steps, extraction of the classical gear residual signal, which is performed by subtracting the time synchronous averaged (TSA) signal from the angular resampled vibration signal, and extraction of the diagnostic features from the filtered gear residual signal using spectral kurtosis and squared envelope. It was shown that the spectral kurtosis based gear fault feature extraction performs effectively, which was validated using simulated and experimental data of a single- and multi-stage gearboxes.
Cranfield also implemented novel three stage decision making technique based on weighted majority rule. This technique is developed to perform tooth wise diagnosis and remove any false alarms produced during the anomaly detection stage.
Experimental vibration data obtained from MODENA and ERIKS were processed using the developed technologies. For both experiments the gearbox is dismantled and visual inspection is carried to identify traces of pitting fault. For MODENA test rigs 1,2 and ERIKs data, clear indication of pitting damage was observed in all teeth and the proposed SK technique could identify the fault clearly. The total probability of correct diagnosis is estimated as 100% using this decision making technology.
Modena commented and reviewed the diagnosis results and performed visual inspection of the pitting damage by dismantling the gearbox assembly.



Cranfield presented a comprehensive state of the art in the signal processing techniques suitable for Root cause analysis. Three groups of techniques proposed to employ are the Short Time Higher Order Chirp Fourier Transform (STHOCFT), the classical and the short time classical bicoherence for processing the stationary and non-stationary gearbox shaft vibration data, and the instantaneous chirp Fourier HOS for processing the vibration data.
The first group of techniques is based on the Short Time Chirp Fourier Transform and the classical and the short time classical bicoherence for processing the gearbox data from stationary and non-stationary speed conditions and detecting imbalance and misalignment of all the gearbox shafts. The technology based on the above mentioned signal processing techniques is developed.
The diagnostic features based on the magnitude of the Short Time Chirp Fourier Transform at first shaft order frequency and magnitudes of the classical and the short time classical bicoherence components involving low shaft orders are recommended considered for the imbalance and misalignment diagnosis in gearboxes. Finally, Cranfield developed the root cause analysis technology based on these recommendations.
For this technology, Cranfield also implemented novel three stage decision making technique based on weighted majority rule. This technique is developed to remove any false alarms. The total probability of correct diagnosis is estimated as 100% using this decision making technology.
The implementation of the developed RCA technology for the imbalance and misalignment detection was demonstrated using the data from the Modena test rig 1 and 2. ISRI reviewed and commented on the RCA technology results.



Cranfield developed the prognosis technology for gears. The dependencies between stress and acceleration are estimated experimentally by Modena by instrumenting the gears with strain gauges. The novel evaluation of FEA stress-vibration dependencies are produced and validated by Modena with the experimentally arrived dependencies. The implemented novel technology involves the following steps: novel detection of the gear resonant frequency and peak amplitude of the gearbox bearing housing acceleration using the short time CFT technique, for the estimated gear resonant frequency and for the measured instantaneous gearbox shaft speed, Estimation of the gear stresses using the FEA stress-acceleration dependencies for multiple mode resonances, Continuous estimation of the gear accumulated fatigue damage factor for multiple mode resonances due to the shaft order harmonic excitations, Continuous estimation of the remaining useful life of rotating units of gearboxes, Decision making about the fatigue damage crack appearance on the gear based on the estimated fatigue damage factor. The technology was tested using Modena data from special experiment with gear resonance.

WP3
The technology (diagnosis, prognosis and root cause analysis) was programmed by Cranfield and delivered by Cranfield to ISRI as a single DLL file and the supporting files (.h and .lib) which contain all the necessary programs for the analysis.
The program was designed for analysis of the recorded data from the gearboxes. The algorithms are developed in MATLAB environment and all the validated MATLAB functions corresponding to the three technologies are placed in a single .m format file. The whole set of functions used by the software can be split into five groups developed based on Cranfield IP back ground, listed as follows,
* Automatic File Detection Technology
* Binary to text conversion
* Diagnosis Technology
* Prognosis Technology
* Root Cause Analysis Technology
The DLL files delivered to ISRI are tested using the experimental gear vibration received from MODENA. The prototype software is based on standard open system architecture software platform that integrates diagnosis, prognosis and root-cause analysis technologies into a single DLL file. The open system architecture software platform according to OSA-CBM 3.1 standard is an implementation of the ISO-13374 functional specifications.
In addition a sixth layer, Root Cause Analysis technology for imbalance and misalignment detection is included in the software. Also Cranfield worked closely with ISRI to solve software integration issues.
ISRI received the DLL program developed by Cranfield and developed the LabView software for IndGear prototype. To integrate the MATLAB generated DLL with the LabVIEW, the DLL is wrapped by an intermediate level of code which does the following:
1. Initialize the MCR (MATLAB Compiler Runtime) and load the MATLAB DLL (the first time it is called).
2. Convert data formats of the driver into MATLAB data (m x Array).
3. Call the MATLAB functions from the MATLAB DLL.
4. Convert MATLAB data back into data type formats that the driver (LabVIEW in our case) can understand.
The software captures data from 4 channels (3 accelerometer channels and 1 channel for speed data), creates data acquisition files which are then passed to the integrated DLL files to perform analysis. The shaft speed can be estimated using the vibration data using the inbuilt Cranfield IP background technology. Results are produced in a file which are then read by the software, interpreted and updates the user with the current status of the gearbox including flagging warnings as necessary. The prototype software creates a file for data acquisition for the duration specified in the acquisition period. Once the duration has passed a new data acquisition file is created. This process keeps going until the stop button is clicked. The DLL files will perform diagnosis; prognosis and root cause analysis on the data and produce a results file.
The group of Modena contributed to the activities of this task, which was concluded at month 15. The contribution of Modena consisted in supporting the development of the platform and in integrating the prognosis algorithms.



ISRI did the following work for hardware selection, design and integration of the prototype components. Hardware equipment, which includes various sensors, a wireless data acquisition system, and a local processing unit for data manipulation and damage diagnosis, prognosis and root cause analysis able to operate under harsh environmental and operating conditions are designed, developed and completed. The equipment is designed to operate without interferences, while being resistant to water and moisture, dust, high or low temperatures, excessive noise and vibrations, etc. In addition, the sensory and data acquisition systems are designed to receive power from a fully independent power supply, in order to avoid any cabling. The proposed sensory, data acquisition and processing system is composed of two independent parts, the remote gearbox part and the local data analysis part.
The local data analysis part includes a computer, a wireless data receiver, and series of data processing algorithms to achieve the desired tasks. As the DAQ system transmits the data wirelessly, data receiver is realized by a wireless router, which is able to communicate with several wireless modules. The router is wired to the computer, which is used for two main purposes: configuration of the wireless DAQ modules and data analysis. In addition, the computer also serves for configuring the parameters of wireless DAQ with the highest importance given to synchronous data sampling. The measurements are transmitted upon analog/digital conversion using IEEE 802.11b/g (Wi-Fi) wireless communication interface. The DAQ system is composed of the data acquisition module for accelerometer (NI-9234), voltage measurements (NI-9215) for shaft speed measurements, and a chassis (NI cDAQ-9191), which is in charge of the wireless data transmission. The NI 9234 includes signal conditioner for ICP type accelerometers. Thus no external antialiasing filter is necessary. This module is placed inside the NI9191 DAQ chassis.
The remote gearbox equipment is placed near to the gearbox, in order to perform measurements of vibration and shaft speed, and transmit data to the local analysis environment.
During hardware design Mikrosay, TEC and Nardoni have provided valuable inputs and professional assessment of the selected components specifically those intended to work in harsh operating environments.



ISRI consulted, ERIKS and NW and all SMEs, made a decision to use wireless communication system. The IndGear hardware and software successfully been integrated into panels built for field testing. The LabVIEW user interface software has also been integrated and deployed on the Human Machine Interface (HMI) touch panel computer. The designed fully-functional physical prototype consists of two panels, one panel is designed to acquire data and transmit data wirelessly to the other panel, which is designed to receive and process the data. As the DAQ system transmits the data wirelessly, data receiver is realized by a wireless router, which is able to communicate with several wireless modules.
Belkin N150 wireless router includes exclusive Multi Beam technology, which minimizes dead spots and provides strong coverage, and the pre-set security settings ensure that the network is safe. With twice the speed of G technology, the N150 is ideal for wireless data transmission, especially when large sampling rates or amount of data is involved. The N150 provides wireless transmission speed up to 150Mbps. The wireless-N technology offers twice the speed of G technology yet is fully compatible with older wireless-G networks, as well as newer 802.11n networks. The panels and its connectors are IP66 rated for weatherproofing and field trials.
The data acquisition panel has been designed and built to operate on either 24VDC or 240VAC depending on the power supply available on site. The HMI panel has been designed to operate at 240VAC. The developed prototype is capable of operating in the harsh environments of wastewater treatment plants.
Tests were carried out to check the communications between the two panels and also test the software running on the Human Machine Interface (HMI) panel. All data log files and results files were checked and compared to the lab testing data to ensure correct operation of the algorithms.
In addition, for each of the channels provided by the DAQ system, a virtual channel was established to enable signal conditioning by means of signal amplification, sampling rates, etc. To check the user interface, the system configuration settings are changed and it is found that the changes were consistent. The sampling frequency is changed and is saved for further data acquisition.
Cranfield collaborated with ISRI to develop the final DLL program with all the three technologies. The DLL files were submitted to ISRI by Cranfield in a particular format suitable for integration of the files to LabView software environment. This format has designed and developed by ISRI.
Modena has provided furnishing expertise in selecting the appropriate hardware to be used in measurement and acquisition. Mikrosay, Nardoni and LUKA have been involved as an advisor in terms of the supporting of the development of the communication system, which would be appropriate for various operating environments. The rest of the consortium has provided valuable inputs and reviews.



ISRI did the following work for hardware and software integration. The IndGear hardware and software have been successfully integrated into panels built for field testing. The LabVIEW user interface software has also been integrated and deployed on the Human Machine Interface (HMI) touch panel computer. The designed fully-functional physical prototype consists of two panels, one panel is designed to acquire data and transmit data wirelessly to the other panel, which is designed to receive and process the data. Due to ISRI liquidation, ISRI did not performed laboratory trials of the prototype. The two panels along with an accelerometer are delivered by ISRI to Cranfield to perform laboratory and field trials using the prototype.
The final prototype delivered by ISRI to Cranfield had few issues to be fixed. Also the task related to testing of the assembled prototype in laboratory conditions has not been performed. Cranfield has fixed the following issues related to the prototype.
1. In the System Configuration Tab, the option to change the time for data acquisition was not active. The prototype could acquire only 25 seconds of data as a default parameter, prior to the correction. To correct this error the Lab View source files were procured from ISRI and the necessary software diagnosis has been performed. As an additional option the software is enhanced to detect changes to the parameters which could be edited using a System_Config.txt file. The parameters now can be changed either directly in GUI or in the pre saved System_Config.txt file.
2. In the DLL Configuration Tab, the option to change sampling frequency and accelerometer sensitivity parameters was also not active. Cranfield successful performed diagnosis and now these parameters can be altered and saved as per requirement.
Cranfield tested the prototype software and hardware in laboratory conditions instead of ISRI. Tests related to wireless communication between the two prototype panels, data acquisition time and proper functioning of the DLL to produce output results file.
Nardoni have been involved as an advisor on the assembled prototype and performing experimental laboratory testing.
Additionally activities involved matching the preliminary gear conditions to technological possibilities of LUKA factory, and comparison of received parameters and final design of the gears – related to constructional demands of test rig2.

WP4
To perform experiments in test-rig conditions, ERIKS has designed a dedicated test rig. Prior to the installation and validation of the prototype, ERIKs along with Cranfield visited Northumbrian water treatment plant to agree upon specifications of gearbox to be tested. To facilitate comparison between test rig and infield test results, ERIKs proposed to design and manufacture two gearboxes with the same specifications. Two brand new, Fenner gearboxes are considered for testing. The first healthy gearbox (gearbox 1) supplied by ERIKS is installed at Northumbrian water plant.
In parallel, the same healthy gearbox (gearbox 2) installed at ERIK’s test facility was tested. The gearbox installed at ERIK’s test facility was subjected to excessive loads which cause natural pitting damage. The damaged gearbox (gearbox 2) is transported to Northumbrian water plant, and then the data acquisition procedure was repeated on the damaged gearbox. After data capture the damaged gearbox (gearbox 2) is replaced with the healthy gearbox (gearbox 1).
Upon Cranfield’s request, ERIKs also modified the gearbox to provide provisions to mount the triaxial accelerometers and speed sensor. The imbalance is introduced at the output shaft using a mounting disk. The imbalance mass is screwed at the end of the mounting disk with the help of screws. Based on ISO 1940-1 (Mechanical vibration – Balance quality requirements for rotor in a constant (rigid) state – Part 1: Specification and verification of balance tolerances), the applied imbalance is calculated as relative percentage of total permissible residual imbalance.
The misalignment is introduced at the input shaft or the motor end. The misalignment effect is developed due to the pull created by the shims attached at the input shaft end. The oil from the gear box is restrained from leaking using special sealant.
Northumbrian water is an active water treatment plant located in UK. For performing prototype validation, Northumbrian water has reserved a part of the machinery under operation. Dedicated staff were made available to provide safety instructions, power connections and installation of Fenner gearbox for testing and prototype validation. The healthy Fenner gearbox was put in routine operation and necessary changes at the site location to accommodate the new gearbox were made.
For prototype installation Cranfield collaborated with ERIKs to finalize the location of accelerometers. To install the speed sensor, (Remote Optical Laser Sensor) for capturing the speed signal, a piece of reflective tape is pasted on the output shaft. A hole is made on the casing covering the output shaft to allow the laser point to focus on the reflective tape.
The Indgear prototype has two panels; the data acquisition panel is placed near the gearbox. A triaxial accelerometer and laser speed sensor are connected to the data acquisition panel. The data is wirelessly received at the data processing panel. The data is captured and the results are verified simultaneously.
Luka has been involved in assisting with installation of the prototype system to the experimental setup at ERIKs, while ensuring appropriate operating environment for the prototype validation to be effective.
Bierens have provided valuable professional input during installation of the experimental setup at ERIKs and have supervised the performance and testing execution.
The group of Modena completed their activities of this task, which was concluded at month 22. Modena continued the experimental testing activities and obtained a full characterization of the test bench.
Mikrosay has been involved during prototype installation providing important professional consulting for the prototype installation to be appropriate and would resemble realistic conditions.



ERIKs designed two gearboxes for testing in laboratory and infield conditions. Laboratory test are performed at ERIKs test facility and infield test are performed at Northumbrian Water treatment plant.
The first gearbox is an undamaged gearbox; this gearbox is used to record vibration under normal conditions and this gearbox is used later as a replacement gearbox to Northumbrian water plant.
At ERIKs test facility three tests were performed to study the technologies and prototype developed.
1. Imbalance tests (RCA technology validation)
2. Misalignment tests (RCA technology validation
3. Natural pitting damage test (Diagnosis technology validation)
Data were captured for four different load and speed combinations. For load combinations full load and reduced load are considered and for speed combinations 90% speed and half speed are considered.
Cranfield systematically planned all the tests performed at ERIKs and Northumbrian Water Plant. In addition to prototype data acquisition, separate portable data acquisition to capture vibration data from two accelerometers is planned. The data are captured using portable data acquisition system consisting of the signal conditioning units (KEMO PocketMaster 1600 active filters and Endevco 133 signal conditioners), connection board SCB 68, data acquisition card NI-6251, portable computer and external hard drive for data storage. The data are captured in digital “raw” format with minimal signal conditioning, stored on the external hard drive and post-processed using the software of the developed technologies. All input signals are low voltage ones (below 24 Volts) therefore no electrical hazards are expected during connection. All inputs of the data capturing system are characterized by high input impedance (>1MOhm) and should not affect existing signal connections.
For the tests performed at ERIKs test facility and Northumbrian water treatment plant the total probability of correct diagnosis was estimated to be 100% using the RCA technology and diagnosis technology.
Cranfield performed tests at ERIKs test facility to capture vibration in normal and damaged conditions. Vibration data with imbalance and misalignment faults is captured and processed using the developed RCA technology.
Northumbrian water supported prototype testing by providing all necessary services at the site location. Dedicated staffs were made available to provide safety instructions, power connections and installation of Fenner gearbox for testing and prototype validation.
Modena continued the experimental testing activities by obtaining a full characterization of the test bench: strain measurement during operation (dynamic conditions); stress vibration dependencies curves; stress-speed dependencies curves; modal analyses for interpretative usage; use of FEA for extrapolate contact pressure from experimental data. All data collected was used for identifying potentiality and limitations of the approaches developed in WP2 and 3. In addition all data collected were used to validate the prototype performance.
LUKA was involved in production and installation of the gear wheels at Modena test rig. LUKA also provided support in estimating the gear wear and pitting effect evaluation after test runs. LUKA planned the future experimental tests and reviwed the results achieved. LUKA redesigned the gear wheels for performing pitting tests and suggested actions to avoid excitation of of resonance frequencies
Bierens and Mikrosay have been involved during the experimental testing and prototype validation, and have provided important advices in experimental testing execution.
The SMEs have provided important advices for optimisation of the integrated system, improvement of the effectiveness and reliability especially under harsh environments.
Due to liquidation of ISRI, Cranfield took responsibility of this task to test the software and hardware of the prototype.. Cranfield procured the LabView source codes from ISRI during the task 3.4 and fixed the software issues,. Later during infield test no issues or bugs were identified.
Cranfield optimized the parameters of the prototype to reduce the computing resources. The software is optimized to store the data acquired for future inspection. The software is also optimized to produce output results into a single results.txt file and all the information related to the data acquired is archived automatically.
Potential Impact:
IMPACT
The IndGEAR project has introduced the radically novel triple technology for damage diagnosis, damage prognosis and root cause analysis of damage through radically novel signal processing of gear vibrations. This means introducing radically new approaches/techniques to processing vibration signals collected from gearboxes while they are in service, as well as using the novel non-linear non-stationary higher order spectra and
spectral correlations and novel non-stationary spectral kurtosis to advance the state of the art in signal processing/vibration condition monitoring.

The estimated IndGEAR system price − made up of the signal processing software objects, data acquisition software, and database, together with the hardware − is around €20,000. Due to the nature of the system, it is straightforward to retrofit and this can be carried out during routine inspections. For this reason, it is expected that initially, the primary market will be retrofit to existing gearboxes. Taking into account that there are already approximately 70,000 wastewater plants operating across Europe, there is a huge opportunity for the retrofit market.

The European water sector is a major economic player that has many positive impacts, both socio-economic and environmental. In recent years, the turnover of this sector (about €50 billion in the EU) grew by an average of 5% per year compared with an average of 2.5% economic growth. In addition, employment in this sector grew faster than turnover, at a rate of between 6 and 7% per year. Thus, there is a huge opportunity for the new market for the IndGEAR technology.

The IndGEAR project will provide its SME participants with specific knowledge, technology and products that will enable them to further differentiate and strengthen their competitive position relative to their competitors. Strategic competitive developments will be achieved in the following key areas:

1. Direct economic business growth through exploitation of the IndGEAR diagnosis and prognosis technology. This will enable us to secure further investment in capital and resources; to improve the companies skill base, our technology portfolio and to invest in increased production capacity

2. New knowledge in the application of condition monitoring technology using higher order spectra, products and services; enabling value-added services and deeper integration within research and development activities

3. Protection of IPR and development of follow-on own innovative products; enabling expansion and further development of the SMEs activities

4. The formation of a key partnership between our companies that embraces key parts of the monitoring and test and inspection supply chain and enables us to exchange key know-how related to our respective domestic sectors that allow our joint venture to gain competitive advantage

5. Enhanced credibility within our sector.

The huge potential market for the IndGEAR monitoring system, both in Europe and globally, will lead to the creation of job opportunities for SMEs in production, installation and maintenance of CBM systems. Within the water industry end-users, increased job opportunities will be created through training in the installation and operation of the new system, both in new water treatment facilities and for retrofit. As this market grows and expands, it will allow us to offer more competitively-priced products for sale. The effect of the project results on standards will be assessed at the major milestones. For rapid exploitation into existing markets, the project scope will be tailored to allow the project results to fall within existing standards. A code of best practice will be established for the manufacture and installation of monitoring equipment to enable its potential to be fully realised.

Machinery failure at wastewater treatment facilities may result in extreme events, such as penetration of hazardous lubricated oil into clean water and serious environmental problems resulting from the release of untreated sewage into rivers and other bodies of water. These events may cause serious health problems for the population. Thus, effective maintenance of machinery is critical to achieving the required water standards.



DISSEMINATION
The consortium has performed an overview of the possibilities of protecting the IPR, mainly focusing on patents and trademark registration. It was found out that the IndGEAR name (as PEBRGIUM INDGEAR OIL, a lubricating fluid for gearboxes) is used by Pebremco (Shell oil company subsidiary in Taiwan), as well as Penrite Oil (an Australian independent international lubricants company) offering 4 products of Indus Indgear oil B220, B320, B460 and MP90. However this has not been trademarked and at the time of writing, the Exploitation
So far, there have been no patents applied, nor trademarks registered.

A website for the project has been created at http://www.indgear.eu/. The website is described in the Deliverable D5.3 report. At present, the site contains 4 placeholders for content but other than a brief introduction to the project objectives and partners, there is as yet no content accessible to outside users. A one-page project flyer has been created, which includes all the relevant information to inform the public about the main features of the project IndGear. At the final stage of the project, the consortium has prepared two videos each approx. of ten and five minute duration. The first video (https://www.youtube.com/watch?v=txfZoFOXrU0) summarizes the IndGear prototype Demonstration at ERIKS laboratory site, where the prototype was employed on a two stage helical gearbox with input speed 1500 rpm and output speed 82 rpm. The second video is related to experimental validation of the IndGear prototype in the Northumbrian Water Treatment Plant (https://www.youtube.com/watch?v=NhnZK-8Pq3g) where the prototype was employed on a two stage helical gearbox with input speed 1500 rpm under undamaged and damaged conditions.
One of the most important events organised by the consortium was The Gear Day (Modena, Italy – 8th March 2016), where Prof. Gelman and Prof. Pellicano presented the main achievements of the project and the technology developed. They made an impressive plenary keynote lectures at the Gear Day event and widely disseminated Indgear results to potential industrial customers from gearbox area. Almost 50 companies with a total number of participants more than 120 including 30 students have participated at the Gear Day event. The event was promoted and summarized also on Modena University website. Prof Gelman made a plenary keynote lecture at the prestige International Conference WESPAC 2015 (Singapore, December 2015) and disseminated IndGear results. In addition, Prof Gelman also made a plenary keynote lecture at the International Conference of Vibration Engineering Society of China (September 2015, China, Shanghai, a plenary keynote lecture at the World Congress of Engineering (July 2015, London, UK, a plenary keynote lecture at the World Congress of Engineering (July 2016, London, UK), and a plenary keynote lecture at the International Conference on Systems and Materials, (Poland, July 2016) and disseminated IndGear results.
Project partners Cranfield (Prof. Gelman) and Modena (Prof. Pellicano) as part of the International Committee of the 13th Intrn’l Conference on Condition Monitoring and Machinery Failure Prevention Technologies (CM 2016/ MFPT 2016) widely disseminated Indgear results at this Conference. Prof Pellicano made an impressive plenary keynote lecture at this Conference and widely disseminated novel FEA and simulation of gearboxes. Prof Gelman and Dr. Chandra made a specialised keynote lecture related to the spectral kurtosis technology for gearboxes. The group of Modena also attended the ICSV23 conference in Greece and partners.

To enable the SME Exploitation Manager and other SME Beneficiaries to “Use” and exploit the Foreground IP, each of the Beneficiaries grants to the SME Exploitation Manager Access Rights for the Background that is “necessary” and “needed” for the exploitation of the Foreground in the Central Applications by the SME Exploitation Manager on a fair and reasonable conditions.
It is agreed that the Foreground IP shall be the property of the SME Participant carrying out the work (which includes both the work carried out by that SME Participant, and the work carried out by the RTD performers which is invoiced to that SME Participant, and paid for in full) generating that Foreground. Further, where several SME Participant(s) have jointly carried out the work generating the Foreground and where their share of the work cannot be ascertained, they shall have joint ownership of such Foreground.
Each Beneficiary is entitled to exploit the Foreground by using, manufacturing and selling products within its business and for this purpose the owning Beneficiary(s) grants to each of the other Beneficiaries Access Rights to the Foreground for exploitation on a royalty free, 5 year, non-exclusive basis limited to “use” in respect of its business. The businesses and “uses” for which Access Rights are granted under this Clause are as follows:
Mikrosay Yazihm ve Elektronik Enerji Sanayi Tic.A.S. will seek to form part of the supply chain with the main focus being the integration and servicing/support of the developed modules, hence, significantly boosting their business in this sector.
Bierens Machinefabreken B.V. will benefit from the exploitation of the IndGEAR system and will be able to offer its customers all the benefits of diagnosis, root cause analysis and prognosis for their gear systems.
Fabryka Maszyn LUKA Sp. Zo.o. will benefit from the exploitation of the IndGEAR system and will be able to offer its customers all the benefits of diagnosis, prognosis and root cause analysis for their gear systems.
Tecnitest Ingenieros SL, the SME Exploitation Manager – project IndGEAR will open up a new market opportunity for Tecnitest, enabling them to perform prognosis, diagnosis and root cause analysis on gearboxes/rotating equipment.
I&T Nardoni SRL have an extended network of collaborative partners for future projects; increased knowledge within vibration condition monitoring technologies for the water industry and commercial benefits through future supply rights of software, hardware, IndGEAR system.


In terms of the demonstration activities, the consortium has all together attended more that 20 conferences and workshops reported in the deliverable D5.5. One of the main demonstration events was organisation of The Gear Day (Modena, Italy – 8th March 2016), where Cranfield and Modena presented the main achievements of the project and the technology developed. They made an impressive plenary keynote lectures at the Gear Day event and widely disseminated Indgear results to potential industrial customers from gearbox area. Almost 50 companies with a total number of participants more than 120 including 30 students have participated at the Gear Day event.
In addition, the consortium has performed several experiments in test rig and in-field conditions (Modena, ERIKS and Northumbrian Waste Water treatment facilities) and documented them in the video form, which can serve as a good basis for demonstration to the public.
List of Websites:
THE PROJECT WEB PORTAL is available to the Project Consortium and the Home Page is also available to the General Public at: http://www.indgear.eu

Name, title and organisation of the scientific representative of the project's coordinator :
Hasan Terzioglu, Managing Director
Mikrosay Yazihm ve Elektronik Enerji Sanayi Tic.A.S.
Tel: +00 90 216 459 8660
Fax: +00 90 216 459 8370
E-mail: hasan.terzioglu@mikrosay.com.tr