HAL Id: hal-01178841
https://hal.archives-ouvertes.fr/hal-01178841
Submitted on 21 Jul 2015
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Distributed under a Creative Commons Attribution - NonCommercial - NoDerivatives| 4.0 International License
Control of heating power in ATP with virtual charts Nicolas Bur, Pierre Joyot, Pierre Villon, Francisco Chinesta
To cite this version:
Nicolas Bur, Pierre Joyot, Pierre Villon, Francisco Chinesta. Control of heating power in ATP with virtual charts. 12e colloque national en calcul des structures, May 2015, Giens, France. �hal-01178841�
Giens 2015
Control of heating power in ATP
with virtual charts
N. Bur
(a,?), P. Joyot
(a), P. Villon
(b), F. Chinesta
(c)(a) ESTIA-Recherche, Technopôle Izarbel, 64210 Bidart, France
(b) UTC, Département GSM, Centre de Recherches de Royallieu, CS 60319, 60203 Compiègne cedex, France (c) GeM Institute, École Centrale de Nantes, 1 rue de la Noë, 44321 Nantes cedex 3, France
(?)n.bur@estia.fr
Problem
Using thermoplastic composites within an Automated Tape Placement process requires controlling drastically the temperature in order to heat the matrix enough to melt it, but not too much to avoid burning. Thus, to reach the optimal temperature within a given zone, we want defining the power of the source by
• integrating in the robot’s control loop • the temperature field
• computed for many values of different parameters
Method
Based on separated representation of the unknown, the Proper Generalised Decomposi-tion (PGD) method allows to handle param-eters as extra-coordinates, leading to virtual charts, without incurring the curse of dimen-sionality.
For example, instead of computing u on a 2D domain (x, z), at each time step (t) and for any value of a parameter (k), we write the unknown under the form
u(x, z, t, k) = X
α
Xα(x)Zα(z)T α(t)Kα(k). Then, assuming first functions are known, the approximation is enriched with a new mode XZT K within a greedy algorithm:
un+1 = n+1X α=1 Xα(x)Zα(z)T α(t)Kα(k) = n X α=1 Xα(x)Zα(z)T α(t)Kα(k) | {z }
known from initialisation or previous computations + + X(x)Z(z)T (t)K(k) | {z } to be computed
# of plies as coordinate
• Unitary domain in the thickness, discre-tised with the Least Common Multiple be-tween the numbers of layers
• Double nodes at each possible interface
• Temperature continuity on fake interfaces (depending on # of plies) enforced with Nitsche’s method 0 1/5 1/4 1/3 2/5 1/2 3/5 2/3 3/4 4/5 1 1 pli C C C C C C C C C 2 plis C C C C RTC C C C C 3 plis C C RTC C C C RTC C C 4 plis C RTC C C RTC C C RTC C 5 plis RTC C C RTC C RTC C C RTC 1 ply 2 plies 4 plies 5 plies
Multi-parametric virtual charts
For a given trajectory, the robot’s velocity is known: we want to determine the best asso-ciated power to ensure the optimal temperature. We assume a static solution is enough when acceleration is small, using a transient model otherwise.
Separating space, # of plies, power and Separating space, # of plies, time, power,
speed: speed and acceleration:
u = P XZP QV u = P XZP T QCV Γ 0 µ− r µ µ + r L 0 qmin q qmax x Φ t v vmin vmax v t v v +γ(t −t0) γ v q q + c(t − t 0) q c t0
Source and velocity parametrisation Power and speed parametrisation
Running the PGD algorithm leads to virtual charts depicted below
0 0.2 0.4 0.6 0.8 1 1.2 2 000 4 000 6 000 8 000 10 000 12 000 Velocity m·s−1 P ow er (W ) uopt = 400 K uopt = 500 K uopt = 600 K
Process window when considering 8 plies
Within the control zone, for a given
p, v and Γ are known. Then q is
com-puted with static chart. The optimum
c is solution of a minimisation problem
0 5 · 10−2 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 300 320 340 360 380 400 420 time of interest optimum determined by minc q (u − uopt)2 time (s) T emp er ature (K) uopt c = 10 000 W·s−1 c = 20 000 W·s−1 c = 28 760 W·s−1 c = 35 000 W·s−1
Choosing the power evolution
This research is part of the Impala project, which is a FUI 11 project, funded by Conseil régional d’Aquitaine.
Numerical results
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 280 300 320 340 360 380 400 uopt ± 5 K Time (s) T emp eratu re (K) V elo cit y× 100 + 280 m ·s − 1 3 000 4 000 5 000 6 000 7 000 8 000 P ow er (W ) Velocity Temperature Temperature – static Power Temperatureevo-lution within the
control zone, using
only static chart
(purple dash-dotted line) or both static and transient (plain green line)