OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
HPC on Cloud for SMEs. The case of bolt tightening.
1. HPC on Cloud for SMEs. The case of
bolt tightening.
A. Gómez, C. Cotelo (CESGA)
G. Rodríguez (Texas Controls), J. Souto (AIMEN)
18/06/2015
2.
3. To provide high performance computing and advanced
communications resources and services to the scientific
community of Galicia and to the Spanish National Research
Council (CSIC), as well as, to institutions and enterprises
with R&D&I activity.
To promote high quality research in Computational
Science in close collaboration with the research community
from Galicia as well as from other regions or countries all
over the world; contributing in this way to the advancement
of science, to transfer technology to industry and
administrations, and as consequence, to the welfare of
society as a whole.
Mission Statement
8. HPC for SMEs
“… HIGHLIGHTS that HPC is a crucial asset for the EU's innovation
capacity and STRESSES its strategic importance to benefit the EU's
industrial capabilities as well as its citizens, by innovating industrial
products and services, increasing competitiveness, and addressing
grand societal and scientific challenges more effectively.”
Draft Council Conclusions on 'High Performance Computing: Europe's place in a
Global Race’” of March 27th, 2013
But for SMEs, HPC does not always mean big supercomputers
9. What is HPC?
High Performance Computing
(HPC) is the use of servers,
clusters, and supercomputers – plus
associated software, tools,
components, storage, and services
– for scientific, engineering, or
analytical tasks that are particularly
intensive in computation, memory
usage, or data management
Intersec360
Process
Conceptual Model
Mathematical Model
Software
Numerical Resolution
Analysis
Validation
New conditions
14. 14 -18-06-2015 Copyright 2014 Members of the Fortissimo Consortium
End user
HPC Service
Provider
Independen
t Software
Vendor
Research
Software
Provider
Application
Expert
HPC Expert
Help with M&S and validation
Vertical Applications/Easy interfaces
Confidentiality
Remote storage/HPC/Visualization
Time-to-solution
What do they need?
FORTISSIMO has received funding from the European Union's
Seventh Framework Programme for research, technological
development and demonstration under grant agreement no
609029
15. Why M&S Tightening?
TEXAS Controls core business.
There is no software solution in the market.
It seems feasible to do it with open source software.
The SME does not have the required computer capacity.
A challenge regarding modelling, simulation, and execution.
15 -18-06-2015 Copyright 2014 Members of the Fortissimo Consortium
FF404: Case of study
FORTISSIMO has received funding from the European Union's Seventh Framework Programme for
research, technological development and demonstration under grant agreement no 609029
17. 17 -Copyright 2014 Members of the Fortissimo Consortium
Tightening examples
Try to Reduce operational time & the time exposure to radiation.
Nuclear plants
18-06-2015
18. Copyright 2014 Members of the Fortissimo Consortium 19 -
The process
(Il processo)
18-06-2015
19. Tightening process
20 -Copyright 2014 - Members of the Fortissimo Consortium
Study and prediction of flanges behavior during tightening process.
What is the best strategy to execute it?
18-06-2015
20. Copyright 2014 Members of the Fortissimo Consortium 21 -
The Model
(Il modello)
18-06-2015
21. 22
Materials
• Flanges
• Gasket
• Bolts
Design
• Nr bolts
• Groove
• Flange
• Bolts
• Gasket
Sequence
parameters
• Initial load
• Residual
load
• Steps of
tightening
• Groups of
tightening
Parametric simulations: huge number of cases to obtain the optimal design
Contact-Friction model
Copyright 2014 Members of the Fortissimo Consortium14-10-2004
The model
24. 25 -Copyright 2014 Members of the Fortissimo Consortium
Parametric jobs: Taguchi
18-06-2015
TAGUCHI METHOD:
The approach for this experiment corresponds to the Taguchi L’16 orthogonal array (up to 4 levels & 5
parameters), i.e. up to 16 parametric jobs . The runtimes for each of these parametric jobs are very similar.
PARAMETERS:
Materials for flanges and gaskets
Number of bolts for each subset (strategy);
Final load value (minimum load ) .
Maximum load for each step.
25. Copyright 2014 Members of the Fortissimo Consortium 26 -
The software
(Il software)
18-06-2015
26. 27 -Copyright 2014 Members of the Fortissimo Consortium
Contact-friction model
18-06-2015
Code_aster (and Abaqus for initial model verification)
Taguchi implementation:
Python script which receives the number of parameters/levels from the web application
and generates the different number of .comm files (input) for the parametric jobs.
Embedded python in the Code_Aster .comm file for controlling the tightening workflow .
Job arrays : same job using different .comm file.
Symmetries, Choices of surfaces main and slaves and Quality of the mesh
Numerical Resolution:
Discrete formulation: ALGO_CONT , ALGO_FRO. Options: LAGRANGIAN or GCP algorithm
Formulation continue: ALGO_RESO_CONT, ALGO_RESO_FROT, ALGO_RESO_GEOM. Options:
POINT_FIX or NEWTON algorithm.
Solvers: MULT_FRONT, MUMPS, LDLT, PCG, PETSC or FETI
In case of MPI solvers: PARALLELISME = CENTRALISE, GROUP_ELEM, MAIL_DISPERSE,
MAIL_CONTIGU, SOUS_DOMAINE
27. 28 -Copyright 2014 Members of the Fortissimo Consortium
Contact-friction model
18-06-2015
Code_aster (and Abaqus for initial model verification)
Taguchi implementation:
Python script which receives the number of parameters/levels from the web application
and generates the different number of .comm files (input) for the parametric jobs.
Embedded python in the Code_Aster .comm file for controlling the tightening workflow .
Job arrays : same job using different .comm file.
Symmetries, Choices of surfaces main and slaves and Quality of the mesh
Numerical Resolution:
Discrete formulation: ALGO_CONT , ALGO_FRO. Options: LAGRANGIAN or GCP algorithm
Formulation continue: ALGO_RESO_CONT, ALGO_RESO_FROT, ALGO_RESO_GEOM. Options:
POINT_FIX or NEWTON algorithm.
Solvers: MULT_FRONT, MUMPS, LDLT, PCG, PETSC or FETI
In case of MPI solvers: PARALLELISME = CENTRALISE, GROUP_ELEM, MAIL_DISPERSE,
MAIL_CONTIGU, SOUS_DOMAINE
28. 29 -18-06-2015 Copyright 2014 Members of the Fortissimo Consortium
HPC – Job arrays (
Taguchi)
Case workflow
34. Values around the groove are critical to
seal the assembly
GUI: Results in critical points
18-06-2015 Copyright 2014 Members of the Fortissimo Consortium
Intermediate results for each tightening step.
Job can be stopped if this results show wrong values.
35. GUI: Jobs monitor
18-06-2015 Copyright 2014 Members of the Fortissimo Consortium
Cancel a parametric job
not the whole simulation
36. GUI: Remote visualization (I)
VNC client through a browser
18-06-2015 Copyright 2014 Members of the Fortissimo Consortium
37. 39 -Copyright 2014 Members of the Fortissimo Consortium
TurboVNC (high-performance version of VNC)
GUI: Remote visualization (II)
18-06-2015
38. Copyright 2014 Members of the Fortissimo Consortium 40 -
Validation
(La validazione)
18-06-2015
39. 1º 2º 3º
Group A 67850N 50880N 50880N
Group B 50880N 50880N 50880N
Case validation: Sensorized Flange
1º 2º 3º
Group A 67850N 50880N 50880N
Group B 50880N 50880N 50880N
Copyright 2014 Members of the Fortissimo Consortium
40. 42 -Copyright 2014 Members of the Fortissimo Consortium
Conclusions
18-06-2015
Configuring the best simulation’s could be a hard work for a no-expert in FEM.
HPC resources are not only needed for simulations which need a high number
of cores each one. Several multicore’s (~16) simulations are needed for running
parametric jobs (up to 16).
Troubles appears very often and they are due to different issues: mesh,
contacts definition, algorithms involved, software issues or MPI libraries.
SMEs need validate their models against real cases.
End-user prefers a GUI application and it must be flexible to incorporate new
functionalities.
Remote visualization is desirable. Best tool to visualize data depends on the
output size.
Value-added services are really needed as FORTISSIMO proposes.
This template is for the WP4xx experiments. The time for the presentation should be 45 minutes, including questions.
2
During the kick-off, the different processes that can be modelled and simulated where
analysed. Finally, the agreement was to model and simulate the most challenged but useful
process within the design and service provision: the tightening.
During the kick-off, the different processes that can be modelled and simulated where
analysed. Finally, the agreement was to model and simulate the most challenged but useful
process within the design and service provision: the tightening.
List the partners and any application software used . Use N/A if not applicable.
Another relevant scope of the project for Texas Controls, have to do with the deep study of the tensional field in flanges. As we mentioned before, typical FEM studies only take into account the final load of the bolts, given by the Engineering Codes. But due to the physical fact that Higher loads are necessary to get residual loads when, risky situations for the components, always happens during the tightening. It is very common in “on site” situations when a flange is leaking, to increase a lot the load applied, producing hidden damages in grooves and bolts. Most of the times, the better solution is to optimize the tightening procedure.
Typical failure of flanges are due fatigue situations when high stress are located in the groves where metallic gaskets are compressed. If high temperatures are involved, then Stress corrosion cracking appears.
So it is necessary to determine the minimum applied load to achieve the maximum sealing force, with no damages of grooves and bolts.
So, as I have just presented so far in the previous slides, one of our most important activities is focused in getting reliable and safe mechanical joints in the industry, so we could summarize this objective in our lemma: “Leak free Plants”
In this slide I would like to explain the problems we have to face day by day in our activity:
In our opinion, Bolted joints, in a Refinery or in a PowerGen plant, should be considered as a welded joint in many aspects. Of course there are not melted materials nor phase changes when we talk about a flanged joint…and, at a first sight, it is more simple to face a temporary joint ( for example a flange) than a welded joint, but the responsibility of the people that tighten a Flange it is the same of a welder. If this joint fails, for example in a pipe of hydrogen in a Hydrocracker, the consecuences could be catastrophic. So we have to know more about the behaviour of the elements of the flanges.
The big problems appears when we have to face the sealing of a non standard flange in critical applications.
Nowadays, Oil Refining Industry have to face problems to refine heavy oils, and, to optimize the production of lighter fractions from heavy oils, (for example tar sands refineries) This this facilities, higher pressure and higher temperatures are necessary.
Even, new technologies for energy generation as Thermal Solar Energy, needs new solutions for new problems. So, more frequently than in previous years, designers are finding solutions far away from the standards, using “ad hoc” designs. Furthermore, the optimization of the desings of Heat Exchangers, Pressure Vessels is mandatory for the Heavy Industry in Europe to better compete and to give value to their customers.
On the other hand, designer usually obviate during the design process, the tightening itself, without knowing that, in many occasions, damages in main parts occurs during the tightening process and not at the final stage of a tightened flanges.
We have detected, then, that engineers usually doesn’t take into account the tightening process itself when designing flanges of pressure vessels, and never considers the time elapsed in such process, thing that in special critical applications or in huge projects should be a critical figure to consider.
In our projects or designs, we constantly apply recommendations of entities such as the British Petroleum Institute, the Fluid Sealing Association, the European Sealing Association, ASME, API, Offshore Operators Association of the United Kingdom (UKOOA) and other research committees, and we usually get involved in mechanical simulations, finite element analysis and empirical testings on how it affects the temperature, thermal cycling, or even cryogenic conditions in the sealing products.
We have studied in many occasions the differences between applied loads and residual loads, the effects of friction when torque is involve to get a correct tightening, etc…Everything lead us to understand the interaction of different elements of the flanged joints to allow us to face the most critical jobs offering guarantees of implementation and operation of no leaking flanged joints.
So, as I have just presented so far in the previous slides, one of our most important activities is focused in getting reliable and safe mechanical joints in the industry, so we could summarize this objective in our lemma: “Leak free Plants”
In this slide I would like to explain the problems we have to face day by day in our activity:
In our opinion, Bolted joints, in a Refinery or in a PowerGen plant, should be considered as a welded joint in many aspects. Of course there are not melted materials nor phase changes when we talk about a flanged joint…and, at a first sight, it is more simple to face a temporary joint ( for example a flange) than a welded joint, but the responsibility of the people that tighten a Flange it is the same of a welder. If this joint fails, for example in a pipe of hydrogen in a Hydrocracker, the consecuences could be catastrophic. So we have to know more about the behaviour of the elements of the flanges.
The big problems appears when we have to face the sealing of a non standard flange in critical applications.
Nowadays, Oil Refining Industry have to face problems to refine heavy oils, and, to optimize the production of lighter fractions from heavy oils, (for example tar sands refineries) This this facilities, higher pressure and higher temperatures are necessary.
Even, new technologies for energy generation as Thermal Solar Energy, needs new solutions for new problems. So, more frequently than in previous years, designers are finding solutions far away from the standards, using “ad hoc” designs. Furthermore, the optimization of the desings of Heat Exchangers, Pressure Vessels is mandatory for the Heavy Industry in Europe to better compete and to give value to their customers.
On the other hand, designer usually obviate during the design process, the tightening itself, without knowing that, in many occasions, damages in main parts occurs during the tightening process and not at the final stage of a tightened flanges.
We have detected, then, that engineers usually doesn’t take into account the tightening process itself when designing flanges of pressure vessels, and never considers the time elapsed in such process, thing that in special critical applications or in huge projects should be a critical figure to consider.
In our projects or designs, we constantly apply recommendations of entities such as the British Petroleum Institute, the Fluid Sealing Association, the European Sealing Association, ASME, API, Offshore Operators Association of the United Kingdom (UKOOA) and other research committees, and we usually get involved in mechanical simulations, finite element analysis and empirical testings on how it affects the temperature, thermal cycling, or even cryogenic conditions in the sealing products.
We have studied in many occasions the differences between applied loads and residual loads, the effects of friction when torque is involve to get a correct tightening, etc…Everything lead us to understand the interaction of different elements of the flanged joints to allow us to face the most critical jobs offering guarantees of implementation and operation of no leaking flanged joints.
So, as I have just presented so far in the previous slides, one of our most important activities is focused in getting reliable and safe mechanical joints in the industry, so we could summarize this objective in our lemma: “Leak free Plants”
In this slide I would like to explain the problems we have to face day by day in our activity:
In our opinion, Bolted joints, in a Refinery or in a PowerGen plant, should be considered as a welded joint in many aspects. Of course there are not melted materials nor phase changes when we talk about a flanged joint…and, at a first sight, it is more simple to face a temporary joint ( for example a flange) than a welded joint, but the responsibility of the people that tighten a Flange it is the same of a welder. If this joint fails, for example in a pipe of hydrogen in a Hydrocracker, the consecuences could be catastrophic. So we have to know more about the behaviour of the elements of the flanges.
The big problems appears when we have to face the sealing of a non standard flange in critical applications.
Nowadays, Oil Refining Industry have to face problems to refine heavy oils, and, to optimize the production of lighter fractions from heavy oils, (for example tar sands refineries) This this facilities, higher pressure and higher temperatures are necessary.
Even, new technologies for energy generation as Thermal Solar Energy, needs new solutions for new problems. So, more frequently than in previous years, designers are finding solutions far away from the standards, using “ad hoc” designs. Furthermore, the optimization of the desings of Heat Exchangers, Pressure Vessels is mandatory for the Heavy Industry in Europe to better compete and to give value to their customers.
On the other hand, designer usually obviate during the design process, the tightening itself, without knowing that, in many occasions, damages in main parts occurs during the tightening process and not at the final stage of a tightened flanges.
We have detected, then, that engineers usually doesn’t take into account the tightening process itself when designing flanges of pressure vessels, and never considers the time elapsed in such process, thing that in special critical applications or in huge projects should be a critical figure to consider.
In our projects or designs, we constantly apply recommendations of entities such as the British Petroleum Institute, the Fluid Sealing Association, the European Sealing Association, ASME, API, Offshore Operators Association of the United Kingdom (UKOOA) and other research committees, and we usually get involved in mechanical simulations, finite element analysis and empirical testings on how it affects the temperature, thermal cycling, or even cryogenic conditions in the sealing products.
We have studied in many occasions the differences between applied loads and residual loads, the effects of friction when torque is involve to get a correct tightening, etc…Everything lead us to understand the interaction of different elements of the flanged joints to allow us to face the most critical jobs offering guarantees of implementation and operation of no leaking flanged joints.
Embebed python in the Code_Aster .comm file (input) for controling the tightening workflow -> Nos permite:
Poner como variables ciertos valores que vendrán del formaulario web (valores para el material, nº de vueltas, nºde tornillos, fuerzas a aplicar)
=> tener una plantilla .comm para usar con cualquier número de tornillos , vueltas…, gracias al uso de variables y bucles python.
Nos permite también acceder a la tabla de desplazamientos que se va escribiendo, tomar el último valor de desplazamiento de un apriete e imponerlo como condición inicial en la siguiente vuelta.
So, as I have just presented so far in the previous slides, one of our most important activities is focused in getting reliable and safe mechanical joints in the industry, so we could summarize this objective in our lemma: “Leak free Plants”
In this slide I would like to explain the problems we have to face day by day in our activity:
In our opinion, Bolted joints, in a Refinery or in a PowerGen plant, should be considered as a welded joint in many aspects. Of course there are not melted materials nor phase changes when we talk about a flanged joint…and, at a first sight, it is more simple to face a temporary joint ( for example a flange) than a welded joint, but the responsibility of the people that tighten a Flange it is the same of a welder. If this joint fails, for example in a pipe of hydrogen in a Hydrocracker, the consecuences could be catastrophic. So we have to know more about the behaviour of the elements of the flanges.
The big problems appears when we have to face the sealing of a non standard flange in critical applications.
Nowadays, Oil Refining Industry have to face problems to refine heavy oils, and, to optimize the production of lighter fractions from heavy oils, (for example tar sands refineries) This this facilities, higher pressure and higher temperatures are necessary.
Even, new technologies for energy generation as Thermal Solar Energy, needs new solutions for new problems. So, more frequently than in previous years, designers are finding solutions far away from the standards, using “ad hoc” designs. Furthermore, the optimization of the desings of Heat Exchangers, Pressure Vessels is mandatory for the Heavy Industry in Europe to better compete and to give value to their customers.
On the other hand, designer usually obviate during the design process, the tightening itself, without knowing that, in many occasions, damages in main parts occurs during the tightening process and not at the final stage of a tightened flanges.
We have detected, then, that engineers usually doesn’t take into account the tightening process itself when designing flanges of pressure vessels, and never considers the time elapsed in such process, thing that in special critical applications or in huge projects should be a critical figure to consider.
In our projects or designs, we constantly apply recommendations of entities such as the British Petroleum Institute, the Fluid Sealing Association, the European Sealing Association, ASME, API, Offshore Operators Association of the United Kingdom (UKOOA) and other research committees, and we usually get involved in mechanical simulations, finite element analysis and empirical testings on how it affects the temperature, thermal cycling, or even cryogenic conditions in the sealing products.
We have studied in many occasions the differences between applied loads and residual loads, the effects of friction when torque is involve to get a correct tightening, etc…Everything lead us to understand the interaction of different elements of the flanged joints to allow us to face the most critical jobs offering guarantees of implementation and operation of no leaking flanged joints.
The critical distances are around the groove, but the values of flexion and torsion of the flanges are important too. It is necessary to control the values around the groove to seal the assembly.
The critical distances are around the groove, but the values of flexion and torsion of the flanges are important too. It is necessary to control the values around the groove to seal the assembly.
VNC client through a browser. This is adequate for small-medium cases (4- 8 bolts) because the outputs are ~ 4GB each parametric job
TurboVNC is a high-performance version of VNC. Big case (24 bolts) generates big outputs > 8.5GB first step
24 bolts: 10h * 16MPI * 8.5GB of output
So, as I have just presented so far in the previous slides, one of our most important activities is focused in getting reliable and safe mechanical joints in the industry, so we could summarize this objective in our lemma: “Leak free Plants”
In this slide I would like to explain the problems we have to face day by day in our activity:
In our opinion, Bolted joints, in a Refinery or in a PowerGen plant, should be considered as a welded joint in many aspects. Of course there are not melted materials nor phase changes when we talk about a flanged joint…and, at a first sight, it is more simple to face a temporary joint ( for example a flange) than a welded joint, but the responsibility of the people that tighten a Flange it is the same of a welder. If this joint fails, for example in a pipe of hydrogen in a Hydrocracker, the consecuences could be catastrophic. So we have to know more about the behaviour of the elements of the flanges.
The big problems appears when we have to face the sealing of a non standard flange in critical applications.
Nowadays, Oil Refining Industry have to face problems to refine heavy oils, and, to optimize the production of lighter fractions from heavy oils, (for example tar sands refineries) This this facilities, higher pressure and higher temperatures are necessary.
Even, new technologies for energy generation as Thermal Solar Energy, needs new solutions for new problems. So, more frequently than in previous years, designers are finding solutions far away from the standards, using “ad hoc” designs. Furthermore, the optimization of the desings of Heat Exchangers, Pressure Vessels is mandatory for the Heavy Industry in Europe to better compete and to give value to their customers.
On the other hand, designer usually obviate during the design process, the tightening itself, without knowing that, in many occasions, damages in main parts occurs during the tightening process and not at the final stage of a tightened flanges.
We have detected, then, that engineers usually doesn’t take into account the tightening process itself when designing flanges of pressure vessels, and never considers the time elapsed in such process, thing that in special critical applications or in huge projects should be a critical figure to consider.
In our projects or designs, we constantly apply recommendations of entities such as the British Petroleum Institute, the Fluid Sealing Association, the European Sealing Association, ASME, API, Offshore Operators Association of the United Kingdom (UKOOA) and other research committees, and we usually get involved in mechanical simulations, finite element analysis and empirical testings on how it affects the temperature, thermal cycling, or even cryogenic conditions in the sealing products.
We have studied in many occasions the differences between applied loads and residual loads, the effects of friction when torque is involve to get a correct tightening, etc…Everything lead us to understand the interaction of different elements of the flanged joints to allow us to face the most critical jobs offering guarantees of implementation and operation of no leaking flanged joints.
The first step showed bigger displacements, the remaining steps involve adjustments over the compression. This behaviour matches with reality, where the last tightening uses smaller force over pre-tensioned material. The distribution of the stress over the flange and gasket showed a higher peak of stress over the groove zone and in the contacts between flange and gasket.
The local compression distance between the flanges determined by the prediction of the numerical model and the experimental test (Figure Local Distance ) was compared. In the graphic it can be seen how the tendency between both methodologies is the same, but the absolute values were different.
The differences are due to geometric tolerances: Positioning and surface texture. The positioning in the FEM model is perfect due to the need for a good surface contact between the flange and the gasket to properly achieve initial values for the contact algorithm. The FEM model sometimes exhibits some minor penetration between parts, which never happens in the experimental tests..
The surface texture implies roughness which is not taken into account in the FEM model. The roughness does not explain 1mm difference, but any defect in the surface of contact could generate that difference. The final results obtained by FEM methodology were validated by TEXAS because the methodology applied and the tendencies obtained were correct. The results from FEM method will be comparatives between different parameters.