Proposal for EC 6FP Integrated Project

Structure of subproject proposal (total length 4-10 pages, clarifying
pictures are welcome)
1.1 Subproject full title: micro-sensor based monitoring and diagnosis for effective on-line risk
management
1.2 Subproject Acronym: MICRO-MONDIAL
1.3 RISKMAN Research Area: RA2. Systems performance monitoring and diagnostics
2.1 Proposing organisation: Tekniker
2.2 Contact person name: Aitor Arnaiz
2.3 Address: Avd. Otaola s/n
2.4 Tel: 34 943 206744
2.5 Fax: 34 943 272757
2.6 Email: [email protected]
2.7 Web site: www.tekniker.es
2.8 Participating organisations and companies, name and country:
REC1
IND2
IND3
IND4
REC5
Type
TEKNIKER
PRUFTECHNIC/ENTEK?
WEARCHECK
(Manufacturing of energy systems)
VTT?/A UK University?
Country
E
D/UK/Other?
D
E/D
FIN/UK
3. Proposal summary (1/3 page)
The lack of reliable and coherent risk management programmes in most industries prevents the application of
current maintenance solutions. Failure to assess maintenance problems effectively are (partially) due to scattered
information, locally distributed and stored on incompatible means, which makes impossible to fully exploit
existing data and results and impede cross-fertilisation of ideas and experiences. Also, there is an absence of an
'holistic' maintenance view, (Combination of different analytical techniques, with integration of condition
monitoring strategies and diagnosis approaches), as well as missing information on alternatives strategies
available for monitoring & diagnosis and their cost-benefit.
An example of the problems that can appear with the absence of ‘holistic’ maintenance strategies is the arrival of
new sensoring systems –based on micro-technology developments-, one of the biggest challenges in maintenance
systems today. In-line/on-line condition monitoring of lubricants and greases, tooling… will be greatly improved
due to the new possibilities offered by the introduction of micro and nanotechnologies in the development of
sensors and the integration of these systems with microelectronics. New properties to be measured, different
technical analysis which implies less quantity of sample (lubricating systems), sizes of sensors up to know
unbelievable… Real time knowledge will be improved and the integration with current knowledge and historical
situations will help to get reliable and coherent risk management programmes.
Good in-line and on-line sensors can be the solution of many unsolved problems such as:
- Accessing to lubricated points in machines working in extreme conditions or in inaccessible engines
- Detecting early stages of degradation and wear, and abnormal performance in lubricants and greases, the
main “root cause” of many severe machine failures.
- Real knowledge of what is happening in the mechanical components in a system
- Establishment of a good predictive and proactive maintenance with the corresponding saving in costs for
the European Industry
However, the integration of new sensoring systems will be very slow if not flexible monitoring strategies exist,
coupled with automated diagnosis and risk-management strategies. The users will stick to old maintenance
strategies, whereas users of new technologies could jeopardise their inversions due to the existence of diagnostic
systems not ready to handle the new on-line information flow.
How can we make the systems flexible and reliable enough? Monitoring-based strategies maximize the use of
data sources, while increasing reliability levels of fault detection on monitored machinery. Comparing to other
strategies, system performance monitoring needs much more information, in order to provide adequate response
to different fault detection sub-problems (Identification of abnormalities, Diagnosis, Prognosis, …) and the
inclusion of information from many different sources (vibrations, temperature, oil, wear debris,…)
Tasks depicted above (monitoring, diagnosis, prognosis, …) are all knowledge-intensive task, as there is an
evident need of expertise to be able to handle them. Thus, in many cases, external consultants indicate the most
effective strategy for cost-benefit maximization and how to interpret the gathered data in order to detect faults on
time. Often, after set-up of monitoring strategies, a software system should remain for helping users in the risk
management assessment tasks. Sometimes, software is intelligent enough as to provide some help in high
analytical features (Severity assessment, diagnosis, prediction, prognosis,…) that can be of real benefit for novel
users, as well as a valuable tool for experienced users in quickly analyzing large volumes of data.
However, to include (and to maintain) such a knowledge system able to infer adequate solutions is not an easy
task. Field knowledge is still scattered, with well known techniques (vibrations) and machinery (electrical motor,
turbines, …), and other less well known (wear debris, oil analysis, machine-tools, elevators). Thus, we may define
the domain knowledge as ‘ill-structured’. That is, there are different degrees of incompleteness and uncertainty in
the causal relation that links the different input symptoms (Vibrations, Lubricants, Wear Debris) and fault
characteristics (Location and type of failure, severity, remaining lifetime, …).
This ‘ill-structure’ is many times due to incompleteness of current knowledge. However, knowledge on the field
grows continually, related to both engineering research and historical records, and will serve to reduce the gaps
that knowledge presents. Eventually, data will grow enough as to constitute a basis for the use of reliability-based
techniques. Given this, it is of prime interest to get the most of the knowledge available.
But how available is the knowledge?. Well, it is normally distributed among different actors as it appears at
following figure, since business (and thus manufacturing) have become worldwide in many cases.
Services & products provider
Technical assistance
Knowledge on machine items (Oil, Bearings, …)
Expertise in special analytical procedures
Non-automated maintenance control
Manual troubleshooting and repair actions
Internet
Machinery builder/
Engineer
Theoretical knowledge on machine
Remote supervision
Machinery statistics
Configuration/Planning
Strategies for diagnosis
and monitoring
Manual monitoring
Extended data analysis
Off-line data collection
Headquarters
Cost benefit analysis
Diagnosis
Fault detection
Severity assessment
Prognosis/Lifetime prediction
Troubleshooting
Machine
Daily monitoring
Off-line data collection
Trobleshooting
Monitoring
Data collection
Abnormality detection
Industrial Plant/ Transport Fleet
Production scheduling/
Maintenance
Plant/Fleet expertise
Communications (MIMOSA, FieldBus, GSM, ...)
As it can be seen, there are several actors that take part in the global maintenance process. Machinery is grouped
into industrial/energy plants/parks (It also could by applied for cargo/passenger fleets), usually in charge of
production systems. Third party consultant and laboratories can handle most of the analytical data not extracted
on-line (Spectral plots, oil analysis, special NDT…). All these actors influence the monitoring and diagnosis
process, that can also be divided into 3 main activities: monitoring, diagnosis (including those fault detection
topics not covered in monitoring phase) and scheduling (with knowledge about what strategy is best: what, when,
where and how to measure)
It is also worthwhile to mention that this distributed environment could serve to overcome two different problems.
The data ownership (The data is ‘distributedly’ owned, so intelligent services have access to the appropriate
allowed ‘views’ of such data) and the data placement (Incredible amounts of data does not need to travel along the
net. Instead, distributed knowledge approaches the data and decides what is important to upload to be analyzed in
a further step –i.e. reliability growth-).
As it have been written at different Semantic web documents, Internet brings a huge amount of data that, if treated
with ‘intelligent tools’, could leverage the use of manufacturing systems, and remove actual limitations on some
e-manufacturing processes. Intelligent web services based on unified ontology approaches should compose these
tools. This system, specifically aimed at overall asset utilisation increase can add real meaning to the already
existing e-manufacturing strategies.
4. Objectives (1/3 page)
In few words, the objective is to design a complete system to perform 3 tasks (configuration, diagnosis,
monitoring). Therefore, in order to develop an acceptable system, the system should cope with:
 An integrated web-based system to support inferential processes to automatically perform previous tasks.
o Some of them preformed locally (tele-diagnosis for monitoring and severity assessment)
o Some of them performed remotely at web-server
 A set of adaptive tools (earning agents) that investigate information sources that can originate changes
into the business knowledge, and perform modifications when needed.
 A study on potential points to be on-line sensorized –and their potential micro-sensor solutionsdepending on different inputs: (the type of access point, the real potential danger of breakdown, the
possibility of taking information from that point., …).
Potential users organizations for such a system can be found at every stage of the maintenance third party service
groups (vibrations, oil-analysis,…) operating worldwide, machine manufacturers, large final users (In areas as
different as machinery producers, transport, energy generation ...)
5. Deliverables (new products, new processes and services, radical innovations; prime deliverable
is expected to be a breakthrough in applicable knowledge to be transferred to industry and
society)
Main deliverable is a prototype distributed software. This can be defined as an intelligent web services software
system for monitoring and diagnosis of manufacturing systems with a distributed schema.
A second deliverable is the definition of a clear methodology (embedded in the software) for unattended
monitoring and maintenance, where traditional distribution of maintenance tasks is completed with a sensoring
task, able to specify the sensoring technologies availables for a given monitoring problem.
6. Justification and potential impact (economic impact, direct and indirect economic benefits,
European dimension, training and education, conformity with EC societal objectives: quality of
life, health, safety, working conditions, employment, and environment)
An effective application of maintenance techniques is a primary concern for Europe if their prominent role in the
world as prime world manufacturer wants to be maintained. Different report sources demonstrate the benefits that
can be extracted from a deeper penetration of predictive maintenance in industrial and civil facilities. For instance,
in 1988, during a survey over 500 plants distributed world-wide (Europe, Canada, Japan and EEUU) that was
conducted by Technology for Energy corporation to identify impact of condition based maintenance on the
economic operation of process and manufacturing industry, it was shown that, after three or more years of
predictive programme application the following facts emerged: Reduction of repair and maintenance costs on 5080%, a 30% increase in revenue and spare parts availability, giving an overall plant profitability increase of 2060%.
However, the incorporation of predictive maintenance strategies is not as fast as desired in Europe. For instance,
in 2000, a DTI maintenance survey showed that, despite efforts made during last decade, only a 35% of
companies use already predictive/proactive maintenance, whereas rest of companies still applies preventive (40%)
and even corrective (25%) alternatives. These figures contrast with those on the Thomas Marketing Information
Centre (1998), where it states that majority of US firms are currently employing some sort of predictive
maintenance.
On the other hand, the main keywords here are “unattended … real-time … Internet”. Some of the worst disasters
come from small faults which haven’t been detected in time. When analysing reasons we found several factors,
sensors don’t collect enough information, or even if information has been properly collected it hasn’t been
monitored in time or properly, or there are not enough sensors due to the cost of them, or continuous monitoring
tasks were to expensive, … If we take the application scope into account its obvious that any improvement in this
area contributes to preserve and enhance the environmental and natural resources, because of the link existing
between machines faults or dysfunctions and natural disasters. We should also point that working like this also
means to save money for the companies, by the improvement of their workers preparation in some concrete
critical subjects to solve the problems in a more rapid and secure way, without risking not specialised workers
life.
Finally, we should also point out that the maintenance of transport systems (marine vessels, road passengers and
goods transport, railway, aircraft) and earthmoving machinery is also a field for application of these technologies,
with small adaptation (since here oil analysis becomes even more relevant) and development of adequate
protocols through mobile communications (GSM, SMS, UMTS).
7. Description of the work (technological approaches and methods, work tasks, their description,
deliverables and work effort in person months) marked according to the following components
(RTD = Research, technological development and innovation-related activities, DEM =
Demonstration activities, TRA = Training)
Some sub-goals are identified below.
1. Identification of maintenance activities (RTD)
A first goal should define what are the tasks to be considered during the project. This definition should bear in
mind the actual status of development on semantic web services, and the new drivers for change in maintenance
strategies (high loads of information flow in both monitoring and diagnosis, changes from repair-focused
strategies to reliability-focused, inclusion of new micro-sensor and wireless technologies for new on-line
monitoring analysis, etc).
A preliminary list of tasks could be as follows:








Sensoring: Identification of potential points
Monitoring: Identification of abnormalities (Excursions from normal operation)
Diagnosis: Assessment of fault severity (incipient, advanced,…)
Diagnosis: Abnormalities analysis for (starting stage of) fault modes recognition. Type and location of the
fault.
Prognosis: Prediction on degradation curve –and residual lifetime- of machine condition given a fault
Diagnosis: Reliability of previous tasks
Troubleshooting: Recommended actions.
Scheduling: Strategies (meta-knowledge) on monitoring and diagnosis.
The identification should result on a complete contextual description (i.e. an organization, agent and task model as
defined in CommonKADS) of the blocks of work . This identification should also bear in mind the two use cases
that will be exploited in work package 5
Starting date: M0; Ending date: M4
Deliverable: Identification of maintenance activities document.
Participants role
Role
REC1 Coordinator
IND2 Requirements concerning systems final deployment and management
IND3 Requirements concerning use of system from third party services
MM
2
1
1
IND4 Requirements concerning end users
REC5
Total
1
1
6
2. Design of multi-agent systems (RTD)
Design of a Multi-agent system architecture should be performed. Here, identified tasks in previous stage should
be should compose the core of intelligent agents (softbots, brokers) that must be placed in a collaborative
environment, where operation is distributed (i.e. monitoring and part of the diagnosis process must be operated
locally).
A vision of characteristics of (finally expected) physical framework (operating environment, languages,…) is also
expected.
Starting date: M5; Ending date: M10
Deliverable: Agent architecture design.
Participants description
Role
REC1 Coordinator (Agent design)
IND2 Support in design
REC5 Agent design
Total
MM
5
1
3
9
3. Design of domain ontology (RTD)
Ontology mapping concerning domain knowledge on this industrial area should be performed. Previous
representation schemes that different participants can have (concerning maintenance, diagnosis, reliability, etc)
should be agreed in order to go ahead. Different representation of knowledge, such as causal networks, rules,
logic,.., could also come from actual research (and industrial) diagnostics systems. The inclusion of multimedia
contents (ferrography images, thermography images, …) can be optional here.
Starting date: M8; Ending date: M13
Deliverable: Ontological domain definition
Participants description
Role
REC1 Coordinator (Domain Ontology mapping)
IND2 Support in design
REC5 Ontology design
Total
MM
5
1
3
9
4. Design of appropriate task ontologies (RTD)
This sub goal should develop the appropriate task ontologies. It is expected that many needed PSM ontologies
will be already built, though some can need to be worked out.
Where possible, existing task ontologies should serve as a basis for each defined task. It could also be expected to
have some input coming from medical applications and from generic developments. Again, actual fault detection
systems developed at engineering research centres will also provide a basis for task ontology development.
Starting date: M8; Ending date: M13
Deliverable: Ontological architecture design
Participants description
Role
REC1 Task Ontology mapping
IND2 Support in design
REC5 Coordinator (Task ontology mapping)
MM
3
1
5
Total
9
5. Deployment/Instantiation on a (some) use case(s) (RTD)
Two use cases are defined
1. On the third-party services provider side, there exist specific laboratory ventures (i.e. oil analysis) that should
offer a corporate side to worldwide companies. Whether samples are analyzed locally (10 different
laboratories) the provision of a unified output is not just a matter of a single interface for all distributed data.
Automated diagnostic systems (some already existing, but of local operation) must cooperate, including
reliability and statistics.
2. On the machinery builder side, some manufacturers (machinery builders, transport…) have a need to reliable
automated monitoring and diagnosis of machines sold to developing countries, in order to assure proper
operation (and minimum maintenance costs) during warrant and maintenance periods. In some cases (energy
generation) there are corporate partners - component builders, machine assemblers, energy exploiters – that
should collaborate in the knowledge sharing.
Test on use cases include the deployment of the complete solution, and testing at how the maintenance strategies
work.
Starting date: M14; Ending date: M22
Deliverable: Results on use case test.
Participants description
Role
REC1 Deployment of use cases
IND2 Coordinator. Deployment of use cases
IND3 End user in use case 1
IND4 End user in use case 2
REC5 Deployment of use cases
Total
MM
5
4
5
5
2
21
6. Dissemination/Exploitation (RTD)
Final –and intermediate- output should be disseminated where appropriate (Workshops, Publications, thematic
networks, …), taking into account multiple sectors that can benefit (Computing, Manufacturing, Engineering,
Energy, Transport,…) .
Starting date: M12; Ending date: M24
Deliverable: Final prototype system
Participants description
Role
REC1 Coordinator dissemination
IND2 Coordinator exploitation
IND3 Exploitation activities
IND4 Exploitation activities
REC5 Dissemination activities
Total
MM
2
2
1
1
1
7
7. Management (RTD)
Management activities include participation in subproject meetings (6 meetings scheduled) as well as the
coordination with riskman.
Starting date: M0; Ending date: M24
Participants description
Role
REC1 Coordinator
IND2
IND3
IND4
REC5
Total
MM
1
0,5
0,5
0,5
0,5
3
8. Partners involved, partner profiles (business idea, size, competence) and the role of each partner
Type Role
REC1 Research
Country
Tekniker (E)
IND2
UK/D/Other
IND3
IND4
REC5
Main Tasks
Coordination
Semantic-web based centralised systems development
Soft/Hard Vendor User requirements. Deployment of final solutions.
Dissemination
Thrid party services User requirements. Final Dissemination
Provider
Machinery
User requirements. Telediagnostics testing. Definition
Manufacturer
Research
Telediagnostics. Remote monitoring
Coordination with otrher activities
D
E
UK/FIN
9. Resources for total subproject and for each partner (resources needed: personnel, equipment
etc; costs, work effort in person months)
Approximate figures are given (in €)
Type
REC1
IND2
IND3
IND4
REC5
Total
Total MM
23
10,5
7,5
7,5
15,5
64
Personnel
207
105
75
75
155
617
Travel
25
20
15
15
20
95
Consumables Durable equip. Total
20
252
10
135
90
90
20
195
60
762
10. Duration (starting date, duration in months)
The project is planned as having a duration of 24 months. M0 to be defined according to global
RISKMAN schedule.
11. Financial plan
A fund of 50% of the total budget cost is requested. The resting 50% will be extracted from participants
own funding schemas.
12. Other issues (e.g. ethical, gender, EC policy related issues)
-