HAL Id: jpa-00223649
https://hal.archives-ouvertes.fr/jpa-00223649
Submitted on 1 Jan 1984
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.
RECENT DEVELOPMENTS IN FIELD AND FORCE COMPUTATION
J. Simkin
To cite this version:
J. Simkin. RECENT DEVELOPMENTS IN FIELD AND FORCE COMPUTATION. Journal de
Physique Colloques, 1984, 45 (C1), pp.C1-851-C1-860. �10.1051/jphyscol:19841174�. �jpa-00223649�
JOURNAL DE PHYSIQUE
Colloque Cl, suppl6ment a u n o I, Tome 45, janvier 1984 page C1-851
RECENT DEVELOPMENTS I N F I E L D AND FORCE COMPUTATION
J. Simkin
Rutherford Appleton Laboratory, Ghilton Didcot, Oxon OX11 OQX, U . K .
RQsum6 - Les progrBs r s c e n t s e n m a t i s r e d e m6thodes d e c a l c u l d e s champs Blectromagn6tiques s o n t p a s s d s e n revue, en s e r d f 6 r a n t e n p a r t i c u l i e r aux t r a v a u x p r d s e n t d s B l a Conf6rence COMPUMAG q u i s ' e s t t e n u e 1 GGnes en j u i n 1983. Les p r 6 c i s i o n s o b t e n u e s 1 l ' a i d e d ' d q u a t i o n s i n t d g r a l e s e t aux
dB-r i v 6 e s p a r t i e l l e s s o n t compardes. Le c a l c u l d e s f o r c e s e s t examin6.
A b s t r a c t - Recent developments i n methods f o r e l e c t r o m a g n e t i c f i e l d compu- t a t i o n a r e reviewed, i n p a r t i c u l a r work r e p o r t e d a t the COMPUMAG Conference h e l d i n Genoa i n June 1983. The accuracy of p a r t i a l d i f f e r e n t i a l and i n t e g r a l e q u a t i o n s o l u t i o n s i s compared and t h e e v a l u a t i o n of f o r c e s i s examined.
1 - INTRODUCTION
The computation of e l e c t r o m a g n e t i c f i e l d s i s important f o r many s c i e n t i f i c and i n d u s t r i a l a p p l i c a t i o n s . Most of t h e s e a p p l i c a t i o n s r e s u l t i n computational problems t h a t a r e unique to t h i s d i s c i p l i n e . One of t h e major problems i s t h a t r e s u l t s must be extremely a c c u r a t e i f they a r e t o be of any p r a c t i c a l u s e . T h i s requirement would be i m p o s s i b l e t o s a t i s f y i f t h e m a t e r i a l s used i n the construc- t i o n of e l e c t r o m a g n e t i c d e v i c e s were l e s s easy t o model. Commonly used m a t e r i a l s a r e e a s i l y measured and c h a r a c t e r i s e d , and t h e i r p r o p e r t i e s a r e r e a s o n a b l y c o n s t a n t . However, t h e r e a r e e x c e p t i o n s , f o r example permanent magnet m a t e r i a l s s u c h a s Alnico. There a r e many o t h e r s p e c i f i c problems, f o r example t h e unbounded n a t u r e of t h e f i e l d , t h e g e o m e t r i c a l complexity of t h e d e v i c e s and s k i n e f f e c t s i n eddy c u r r e n t s o l u t i o n s .
T h i s paper reviews t h e p r o g r e s s i n e l e c t r o m a g n e t i c f i e l d computation, paying p a r t i c u l a r a t t e n t i o n t o work r e p o r t e d a t t h e June 1983 COMPUMAG conference.
C u r r e n t r e s e a r c h i s dominated by t h e a p p l i c a t i o n of f i n i t e elements to t h e s o l u t i o n of p a r t i a l d i f f e r e n t i a l e q u a t i o n s , t h e reasons f o r f a v o u r i n g t h i s approach a r e examined. The c a l c u l a t i o n of f o r c e s produced by e l e c t r o m a g n e t i c f i e l d s i s o f t e n s u b j e c t t o l a r g e e r r o r s , t h e t e c h n i q u e s used f o r f o r c e c a l c u l a t i o n s a r e compared and the cause of the e r r o r s i s i d e n t i f i e d .
2
- ACCURACY
I n many a p p l i c a t i o n s of e l e c t r o m a g n e t i c d e v i c e s t h e f i e l d s must be computed t o an a c c u r a c y of t h e o r d e r of 1 p a r t i n 1000 o r b e t t e r . T h i s can only be achieved when t h e m a t e r i a l s involved a r e s t a b l e and t h e i r p r o p e r t i e s can be e a s i l y measured.
Even t h e n , g u a r a n t e e i n g t h e r e s u l t s t o t h i s accuracy i s extremely expensive and f o r t r u l y t h r e e dimensional f i e l d s may be i m p o s s i b l e . Problems t h a t i n v o l v e l i n e a r magnetic m a t e r i a l s can be s o l v e d t o v e r y high p r e c i s i o n i n two dimensions u s i n g boundary i n t e g r a l m e t h o d s / l / . A s i m p l e problem has been solved w i t h c o n s t a n t p e r m e a b i l i t y i n o r d e r t o show r a t e s of convergence of t h e s o l u t i o n f o r d i f f e r e n t methods. Although r e l a t i v e l y simple the problem shown i n F i g u r e 1 does provide a n o n - t r i v i a l t e s t . One q u a r t e r of a 'H' frame magnet with a s l o p i n g p o l e was s o l v e d
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:19841174
Cl-852 JOURNAL DE PHYSIQUE
using a s c a l a r p o t e n t i a l boundary i n t e g r a l method and a v e c t o r p o t e n t i a l , p a r t i a l d i f f e r e n t i a l e q u a t i o n , f i n i t e element method. The problem was assumed i n f i n i t e l y long out of the p l a n e , with an i r o n r e l a t i v e p e r m e a b i l i t y of 1000. A v e r y l a r g e (500 degrees of freedom) boundary i n t e g r a l s o l u t i o n was used t o o b t a i n t h e ' c o r r e c t ' r e s u l t . F i g u r e 2 shows the f l u x d i s t r i b u t i o n i n the problem and from t h i s f i g u r e t h e boundary c o n d i t i o n s should be obvious. F i g u r e s 3a and 3b show t h e convergence of the i n t e g r a t e d f l u x a c r o s s the a i r gap f o r the two s o l u t i o n methods and a l s o show t h e time required f o r the s o l u t i o n s . This problem was s e l e c t e d because the r a p i d l y varying f i e l d gives a more r e a l i s t i c t e s t of t h e f i n i t e element method, accuracy i s much e a s i e r t o o b t a i n i f t h e f i e l d i s c o n s t a n t . I n both methods q u a d r a t i c v a r i a t i o n of t h e p o t e n t i a l was used i n each d i s c r e t e element.
The p a r t i a l d i f f e r e n t i a l equation s o l u t i o n employed smoothing on t h e f i e l d s o l u t i o n , d i r e c t d i f f e r e n t i a t i o n of the element shape f u n c t i o n s g i v e s much l a r g e r e r r o r s .
F i g u r e 1. Cross s e c t i o n of an H frame magnet with a s l o p i n g pole. Only one q u a r t e r of t h e magnet is d i s p l a y e d .
F i g u r e 2. Flux d i s t r i b u t i o n i n t h e H frame magnet.
The r e s u l t s c l e a r l y show t h a t f o r t h i s problem t h e boundary i n t e g r a l method is by f a r t h e b e s t . However, t h i s cannot be used f o r non-linear problems except by i n t r o d u c i n g a r e a d i s c r e t i s a t i o n , whereas rhe p a r t i a l d i f f e r e n t i a l e q u a t i o n method can be used f o r non-linear s o l u t i o n s and w i l l g i v e s i m i l a r p r e c i s i o n . The f i g u r e s do not t e l l t h e whole s t o r y , the i n c r e a s e i n accuracy from 0.08% t o 0.02% with p a r t i a l d i f f e r e n t i a l equatioh s o l u t i o n was obtained by only i n c r e a s i n g t h e
d i s c r e t i s a t i o n w i t h i n t h e a p e r t u r e of the magnet. This s t r o n g c o r r e l a t i o n between
accuracy and $rely l o c a l s u b d i v i s i o n i s one of t h e s i r e n g t h s . o f t h e method, i t i s
not a property of the i n t e g r a l method except i n a r e a s c l o s e to the d i s c r e t i s e d
Error
O/O0 100 200 300 0 2000 4000 6000 8000 Degrees of Freedom Degrees of Freed om
CPU seconds CPU seconds
Fig. 3 A Fig. 38
10
Error in / B dx as a function of Discretisation
0 Y
F i g u r e 3. R e s u l t s f o r t h e H frame magnet.
a ) from a boundary i n t e g r a l s o l u t i o n
b) f r a n a p a r t i a l d i f f e r e n t i a l e q u a t i o n s o l u t i o n 2.1 Cost Comparison
The r e s u l t s p l o t t e d i n F i g u r e 3 were o b t a i n e d by programs t h a t g i v e n e a r l y optimal performance f o r t h e t y p e s of method they use. Table 2 shows how t h e o p e r a t i o n count v a r i e s a s a f u n c t i o n of accuracy f o r boundary i n t e g r a l , volume i n t e g r a l and p a r t i a l d i f f e r e n t i a l e q u a t i o n methods i n 2D s o l u t i o n s . Accuracy i s assumed t o depend on t h e element s i d e l e n g t h r a i s e d t o a power t h a t i s independent of the method. Table 3 shows t h e same comparison f o r 3 D s o l u t i o n s .
Table
2A comparison o f ' o p e r a t i o n counts f o r d i f f e r e n t s o l u t i o n methods i n 2D problems
Boundary Volume P a r t i a l
i n t e g r a l i n t e g r a l D i f f e r e n t i a l
a c c u r a c y
ana
d i s c r e t i s a t i o n
Qn
m a t r i x e v a l u a t i o n
an 2
e q u a t i o n s o l u t i o n a n 3
f i e l d r e c o v e r y n
JOURNAL DE PHYSIQUE
Table 3
A comparison of o p e r a t i o n counts f o r d i f f e r e n t s o l u t i o n methods i n 3D problems
Boundary Volume P a r t i a l
i n t e g r a l i n t e g r a l D i f f e r e n t i a l
a c c u r a c y
an
ad i s c r e t i s a t i o n a n 2 m a t r i x e v a l u a t i o n a n 4 e q u a t i o n s o l u t i o n a n 6 f i e l d recovery a n2
The most obvious c o n c l u s i o n from t h e s e t a b l e s i s t h a t i n t h e l i m i t of v e r y l a r g e d i s c r e t i s a t i o n s all methods a r e dominated by e q u a t i o n s o l u t i o n . The next
c o n c l u s i o n is t h a t f o r s o l u t i o n s i n t h r e e dimensions p a r t i a l d i f f e r e n t i a l methods must become more e f f e c t i v e than t h e o t h e r approaches; a g a i n , i n t h e l i m i t of l a r g e d i s c r e t i s a t i o n s when t h e c o n s t a n t s i n t h e p r o p o r t i o n a l i t y r e l a t i o n s h i p s become i n s i g n i f i c a n t . It is only i n two dimensions t h a t boundary i n t e g r a l methods a r e p a r t i c u l a r l y a t t r a c t i v e , F i g u r e 3 d e m o n s t r a t e s t h i s q u i t e e f f e c t i v e l y . I n s o l v i n g p r a c t i c a l problems on e x i s t i n g computing hardware t h e c o n s t a n t s i n the propor- t i o n a l i t y r e l a t i o n s h i p s become i m p o r t a n t . Experience with e x i s t i n g programs i n t h r e e dimensions shows t h a t once t h e number of unknowns r i s e s above 1000 f o r i n t e g r a l methods, t h e p a r t i a l d i f f e r e n t i a l methods a r e b e t t e r . Accuracies of t h e o r d e r of .2% can be achieved w i t h t h i s l e v e l of d i s c r e t i s a t i o n .
2.2 L i n e a r Algebra
I n d e r i v i n g t h e o p e r a t i o n c o u n t s f o r T a b l e s 2 and 3 i t was assumed t h a t d i r e c t e q u a t i o n s o l u t i o n methods were used f o r i n t e g r a l methods (Gaussian e l i m i n a t i o n ) and t h a t the most e f f e c t i v e methods a v a i l a b l e were used f o r t h e p a r t i a l d i f f e r e n t i a l methods. I n t h e l a t t e r , t h e r e have been s i g n i f i c a n t developments i n t h e l a s t 5
y e a r s and i t is o n l y a s a r e s u l t of t h e s e t h a t the p a r t i a l d i f f e r e n t i a l methods have been s o s u c c e s s f u l . The m a j o r i t y of new, l a r g e , f i n i t e element programs now u s e p r e c o n d i t i o n e d con j u g a t e g r a d i e n t methods/2,3/. The computer s t o r a g e r e q u i r e d by t h e s e methods i n c r e a s e s l i n e a r l y w i t h t h e number of unknowns and i s independent of o r d e r i n g of t h e unknowns. S i m i l a r l y t h e s o l u t i o n times a r e almost independent of t h e o r d e r i n g of t h e unknowns, t h i s i s t o be c o n t r a s t e d with o t h e r s p a r s e m a t r i x methods where t h e bandwidth o r p r o f i l e of t h e m a t r i x was v e r y important e s p e c i a l l y f o r problems i n t h r e e dimensions. Table 4 g i v e s an i n d i c a t i o n of t h e s o l u t i o n t i m e s r e q u i r e d u s i n g an incomplete cholesky c o n j u g a t e g r a d i e n t method f o r a f i n i t e element s o l u t i o n o f L a p l a c e ' s e q u a t i o n i n t h r e e dimensions.
Table 4
S o l u t i o n times f o r incomplete cholesky c o n j u g a t e g r a d i e n t methods.
An I B M 3081D was used f o r t h e s e c a l c u l a t i o n s .
Number of S o l u t i o n
unknowns time ( s e c o n d s )
The R u t h e r f o r d Appleton Laboratory implementations of t h e s e methods do not use any d i s k backing s t o r e , t h e s o l u t i o n of a 25000 unknown s e t of e q u a t i o n s f o r t h e L a p l a c i a n problem r e q u i r e s 5MB of v i r t u a l s t o r a g e .
2.3 3D S o l u t i o n s
I n o r d e r t o show the convergence of s o l u t i o n s t o t h r e e dimensional problems an
axisymmetric model w i t h t h e c r o s s - s e c t i o n
as shown i n F i g u r e 1 was solved using a2D a x i s y m e t r i c model and a f u l l 3 D model i n which one o c t a n t of t h e geometry was d i s c r e t i s e d . F i g u r e 4 shows a computer generated d i s p l a y of the model with hidden l i n e s removed. F i g u r e 5 shows t h e convergence of t h e i n t e g r a t e d f i e l d a s a f u n c t i o n of d i s c r e t i s a t i o n . Moving from F i g u r e 3 t o F i g u r e 5 using Tables 2 and 3 i t would seem reasonable t o expect t o r e q u i r e 600000 degrees of freedom t o g i v e 0.2% accuracy. Figure 5 shows t h a t only 24000 unknowns were r e q u i r e d , t h i s was achieved by using high o r d e r small elements i n t h e r e g i o n of i n t e r e s t and low o r d e r elements i n t h e o u t e r a r e a s . These r e s u l t s were obtained using the TOSCA
F i g u r e 4. A 3 D model of an axisymmetric H frame magnet.
E r r o r %
I I I