<?xml version="1.0" encoding="UTF-8"?><!DOCTYPE article  PUBLIC "-//NLM//DTD Journal Publishing DTD v3.0 20080202//EN" "http://dtd.nlm.nih.gov/publishing/3.0/journalpublishing3.dtd"><article xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" dtd-version="3.0" xml:lang="en" article-type="research article"><front><journal-meta><journal-id journal-id-type="publisher-id">WJM</journal-id><journal-title-group><journal-title>World Journal of Mechanics</journal-title></journal-title-group><issn pub-type="epub">2160-049X</issn><publisher><publisher-name>Scientific Research Publishing</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="doi">10.4236/wjm.2011.13015</article-id><article-id pub-id-type="publisher-id">WJM-5774</article-id><article-categories><subj-group subj-group-type="heading"><subject>Articles</subject></subj-group><subj-group subj-group-type="Discipline-v2"><subject>Engineering</subject><subject> Physics&amp;Mathematics</subject></subj-group></article-categories><title-group><article-title>
 
 
  Asymptotic Expansion of Temperature Close to a Singularity of a Plate
 
</article-title></title-group><contrib-group><contrib contrib-type="author" xlink:type="simple"><name name-style="western"><surname>sabelle</surname><given-names>Titeux</given-names></name><xref ref-type="corresp" rid="cor1"><sup>*</sup></xref></contrib></contrib-group><author-notes><corresp id="cor1">* E-mail:<email>isabelle.titeux@univ-reims.fr</email></corresp></author-notes><pub-date pub-type="epub"><day>29</day><month>06</month><year>2011</year></pub-date><volume>01</volume><issue>03</issue><fpage>109</fpage><lpage>114</lpage><history><date date-type="received"><day>March</day>	<month>22,</month>	<year>2011</year></date><date date-type="rev-recd"><day>April</day>	<month>21,</month>	<year>2011</year>	</date><date date-type="accepted"><day>May</day>	<month>5,</month>	<year>2011</year></date></history><permissions><copyright-statement>&#169; Copyright  2014 by authors and Scientific Research Publishing Inc. </copyright-statement><copyright-year>2014</copyright-year><license><license-p>This work is licensed under the Creative Commons Attribution International License (CC BY). http://creativecommons.org/licenses/by/4.0/</license-p></license></permissions><abstract><p>
 
 
  The thermal conduction in a thin laminated plate is considered here. The lateral surface of the plate is not regular. Consequently, the boundary of the middle plane admits a geometrical singularity. Close to the origin, the lateral edge forms an angle. We shall prove that the classical bidimensional problem associated with the thin plate problem is not valid. In this paper, using the boundary layer theory, we describe the local behavior of the plate, close to the perturbation.
 
</p></abstract><kwd-group><kwd>Thin Plate</kwd><kwd> Boundary Layers</kwd><kwd> Singularities</kwd><kwd> Asymptotic Expansion</kwd></kwd-group></article-meta></front><body><sec id="s1"><title>1. Introduction</title><p>A thin plate is a three dimensional body, a dimension of which (the thickness) is smaller than the other dimensions. Usually, under the assumption of small thickness with respect to the characteristic length of the middle plane, instead of a three-dimensional description, a bi-dimensional one is used. The problem is then posed over the middle plane of the plate. In this way, the numerical methods are less expensive in time and in memory.</p><p>In this paper, we deal with the modelling of the thermal behavior of a thin laminated plate. The thermal conductivity can be considered as an exemplary problem similar to the elasticity problem. Instead of displacement, the unknown is the temperature, but the equations are similar.</p><p>The temperature and thermal flux established by the asymptotic expansion are good approximations. But close to the lateral surface, the bi-dimensional behavior is not suited: for a laminated body, boundary conditions are only satisfied on average. However, on the edge damage phenomena can appear, like delamination, crack... A bi-dimensional description of the behavior of the plate is not enough. In order to predict these phenomena, we need a good, local description of the behavior of the plate, in these areas. In this case, the bi-dimensional expansion is no more sufficient, we need a local three-dimensional description, which is valid only close to the lateral surface. Moreover, the distance to the edge and the position in the thickness have the same range of magnitude; the assumption of small thickness with respect to the other directions is no more valid. Close to the edge, the body must be considered like a threedimensional domain, with thickness and distance to the edge of the same order.</p><p>In previous works, the cases of a classical regular edge [<xref ref-type="bibr" rid="scirp.5774-ref1">1</xref>] and edge with a local perturbation [<xref ref-type="bibr" rid="scirp.5774-ref2">2</xref>] were considered. Recently, Saidi et al. [3,4] studied singularities in the edge of moderately thick plates. They studied the effect of the boundary layer term added to the Mindlin’s plate theory. In this paper, the case of thin laminated plate with an angle in the lateral edge is considered. We lose the symmetry of the local perturbation [<xref ref-type="bibr" rid="scirp.5774-ref2">2</xref>] and new arguments must be used. Because of this angle, the boundary of the plate cannot be considered as a smooth surface. The boundary layer theory must be adapted to take into account the new geometry of the perturbed edge. Therefore, we obtain a new local description which is posed on an unbounded domain. The existence and uniqueness of the solution must be proved in order to implement a numerical method of resolution.</p></sec><sec id="s2"><title>2. Generalities and Description of a Thin Laminated Plate</title><sec id="s2_1"><title>2.1 Classical Problem for a Thin Plate</title><p>Insert <xref ref-type="fig" rid="fig1">Figure 1</xref>: The plate <img src="6-4900022\1cdbd185-e66d-4baf-b562-e4748072cb9c.jpg" /></p><p>Let us consider a thin plate <img src="6-4900022\58ce552b-df26-435e-bf75-c518c3b42912.jpg" /> characterized by its middle plane <img src="6-4900022\242c4213-0063-402b-8a8a-e7af8f898f02.jpg" /> &#160;and its thickness</p><p><img src="6-4900022\7f5858f1-a828-4118-a554-dc8dbb8eefbd.jpg" />(cf. <xref ref-type="fig" rid="fig1">Figure 1</xref>). In fact, <img src="6-4900022\e60859a8-2e2c-4cff-bda2-ee190b369125.jpg" /> denotes the ratio between the characteristic length of the middle plane of the plate and the thickness. Let <img src="6-4900022\a66005ef-8457-4174-bc04-775dfe8d64aa.jpg" /> and <img src="6-4900022\b7373f37-8acd-46f1-b34a-54a51c45fbc2.jpg" /> denote the upper and lower faces respectively and <img src="6-4900022\e816a4f2-a325-4f02-8d65-0ed5be43ba44.jpg" /> denotes the lateral edge.</p><p>Coordinates in the middle plane are <img src="6-4900022\097823a6-13e9-44e4-a4c4-608e8d66bef8.jpg" /> and position in the thickness is<img src="6-4900022\78f66a42-e644-45cc-96a6-752911098eb1.jpg" />. Symbols in boldface denote vectors. We use the summation convention on repeated indices. Lati<img src="6-4900022\0d618e96-48cd-4a16-b605-5674fc130088.jpg" />n indices take their values in the set <img src="6-4900022\cd594468-ab3b-432c-aadb-33752f9781f1.jpg" /> while Greek ones take their values in the set<img src="6-4900022\bdbba478-a76b-4ed7-80f9-70b90f76e1a8.jpg" />.</p><p>For given external sources of heat, we have to determine the temperature field<img src="6-4900022\e958a168-e80d-4143-ab91-ce99054370b2.jpg" />. The thermal flux vector is related to <img src="6-4900022\a7680e19-ada0-44a0-a203-d16bdc51b52b.jpg" /> by the constitutive law</p><p>&#160; <img src="6-4900022\0f7a97e3-1f43-4919-903f-0df64fa2a328.jpg" /></p><p>The coefficients<img src="6-4900022\46c81ee7-2e4a-4236-9f76-5552213fe621.jpg" /> are the conductivity coefficients. They satisfy symmetry and coerciveness properties:</p><p><img src="6-4900022\d6248165-5e0a-4775-935a-ac50700f3c32.jpg" /></p><p><img src="6-4900022\54024421-3b3a-4bda-8038-ecccc13ad715.jpg" /></p><p>The equilibrium equation is</p><disp-formula id="scirp.5774-formula123013"><label>(2.1)</label><graphic position="anchor" xlink:href="6-4900022\01106030-db6b-43a6-a2fd-475198208d57.jpg"  xlink:type="simple"/></disp-formula><p>The order of magnitude of <img src="6-4900022\5a946bbe-4451-4ef5-a9f9-3d21c1662695.jpg" /> is <img src="6-4900022\0e0cb250-f6dd-44c3-922b-2f75e9febaba.jpg" /> it means that <img src="6-4900022\1b2f11b9-2d4d-4b59-8a34-0bb405cb448c.jpg" /> has the same order than the characteristic length of the middle plane. The upper and lower faces of the plate are free of heat source</p><disp-formula id="scirp.5774-formula123014"><label>(2.2)</label><graphic position="anchor" xlink:href="6-4900022\ac503f5f-025d-47a5-935c-5f5405692371.jpg"  xlink:type="simple"/></disp-formula><p>On the lateral edge of the plate, there are Neumann's boundary conditions:</p><disp-formula id="scirp.5774-formula123015"><label>(2.3)</label><graphic position="anchor" xlink:href="6-4900022\0e128bc5-da4e-4350-bc56-8f04e0c52c1a.jpg"  xlink:type="simple"/></disp-formula><p>where <img src="6-4900022\a5144b23-ed9e-49f6-8208-4205f4d76c8f.jpg" /> is <img src="6-4900022\22161289-139b-412e-81ee-f93ca6848ba9.jpg" /> and n is the outer normal.</p><p>The external sources of heat satisfy the compatibility condition</p><p><img src="6-4900022\c62171dd-373e-4391-b4c1-a261a8fb76db.jpg" /></p><p>The plate is laminated, i.e. composed of several materials. We assume that the interfaces between two materials are parallel to the middle plane of the plate. In this way, the conductivity coefficients depend on the position in the thickness<img src="6-4900022\289d2aac-e70b-48a3-a384-0234443d3968.jpg" />, we assume that they do not depend on the other variables:</p><p><img src="6-4900022\cdea24e5-f0d2-4af7-8116-a1fc8f3afc07.jpg" /></p><p>At the interface between two different materials, the temperature and the normal thermal flux are continuous:</p><disp-formula id="scirp.5774-formula123016"><label>(2.4)</label><graphic position="anchor" xlink:href="6-4900022\32a53eaf-d560-4b02-8cff-6c1e31e856a8.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123017"><label>(2.5)</label><graphic position="anchor" xlink:href="6-4900022\72b3f7f9-b4f2-4517-95c9-1edb99ac73b3.jpg"  xlink:type="simple"/></disp-formula><p>where the brackets denote the jump across the interfaces.</p><p>Problem (2.1)-(2.5) is the plate problem.</p><p>Remark 1. In (2.1) and (2.3), external sources of heat are taken with an order of magnitude of 1 in order to get an asymptotic expansion of the solution with the leading term <img src="6-4900022\59db882e-7499-4d17-860d-f292a43791b0.jpg" /> (see (2.7) thereafter). But, because of the linearity of the problem, if the external sources of heat are multiplied by a constant (even depending on<img src="6-4900022\0db09c5d-5031-423a-a664-d375fec55f4e.jpg" />), the solution is also multiplied by the same constant.</p><p>Using the change of variables</p><disp-formula id="scirp.5774-formula123018"><label>(2.6)</label><graphic position="anchor" xlink:href="6-4900022\36403aa4-917f-49d8-a332-deeb870589ec.jpg"  xlink:type="simple"/></disp-formula><p>the asymptotic expansion theory [5,6] involves temperature of the form</p><disp-formula id="scirp.5774-formula123019"><label>(2.7)</label><graphic position="anchor" xlink:href="6-4900022\75ff05db-0fad-4921-93c2-aea8ce017601.jpg"  xlink:type="simple"/></disp-formula><p>The functions <img src="6-4900022\0a696182-5825-412c-aeaa-f96d07ae7368.jpg" /> only depend on the conductivity coefficients; they are solutions of the variational problems</p><p><img src="6-4900022\034d6f38-e715-4c83-92c8-7c80b77cf955.jpg" /></p><p><img src="6-4900022\74d104f8-fbf2-4dbb-a078-f1d01fa2548c.jpg" /></p><p>where<img src="6-4900022\2a87c1bc-0415-4f0a-a019-48aa1581948e.jpg" />.</p><p>The change of variables (2.6) is equivalent to dilate the thickness of the plate. In this way, we obtain a new plate <img src="6-4900022\65b5085e-156b-4b38-8945-2d01ed610824.jpg" /> which does not depend on <img src="6-4900022\97c854b1-7502-439b-8cbc-0e33b312d203.jpg" /> (cf. <xref ref-type="fig" rid="fig2">Figure 2</xref>). All the directions of the plate have now the same range of magnitude. The thickness of <img src="6-4900022\8903a2e0-f3dd-4af0-b49c-4659013518b5.jpg" /> is no more small with respect to the other directions. We shall denote by</p><p><img src="6-4900022\f9134c72-3e49-4c74-82f0-bba302d8ebb8.jpg" />and <img src="6-4900022\04118f89-3366-48a0-b74d-60dc343e2b8b.jpg" /> the upper and lower faces of the plate<img src="6-4900022\f6d02193-e521-49e3-b85f-2fe0accdce13.jpg" />, and by <img src="6-4900022\c6eff91e-b694-4fc8-b01e-7c54d8be807f.jpg" /> the lateral edge.</p><p>Insert <xref ref-type="fig" rid="fig2">Figure 2</xref> The plate<img src="6-4900022\90110196-2401-4c1a-b72c-705470d0d07b.jpg" />.</p><p>Let us remind certain features of the asymptotic structure of the plate problem [5,6]. The asymptotic structure of the mean value of the thermal flux is of the form:</p><p><img src="6-4900022\84794332-bf9d-4e02-9de9-e3fc6303747b.jpg" /></p><p>where the tilde denotes the average on the thickness and <img src="6-4900022\6d685086-a981-4b38-91a4-889bc27c227f.jpg" /> is the leading term of the thermal flux. The homogenized conductivity coefficients are given by</p><p><img src="6-4900022\7d700925-54d6-4027-8fd3-05ff4bb34bd8.jpg" />,</p><p><img src="6-4900022\6e4bdb2c-9bf7-41d4-a1b7-607ee41524e9.jpg" />The problem for<img src="6-4900022\7658649a-3780-49b7-a860-9c7b15cd513f.jpg" /> is posed over the middle plane. So that it is a bi-dimensional problem</p><disp-formula id="scirp.5774-formula123020"><label>(2.8)</label><graphic position="anchor" xlink:href="6-4900022\0441bc12-8b46-41b0-b9e1-f43afabd13d9.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123021"><label>(2.9)</label><graphic position="anchor" xlink:href="6-4900022\a83ad076-6234-4f9c-9fce-9c0a0d8e6cb3.jpg"  xlink:type="simple"/></disp-formula><p>where <img src="6-4900022\18b747dd-aa55-4633-a8df-3ad883057bb4.jpg" /> and <img src="6-4900022\263208ca-c7e8-4039-ab81-ae056c28098c.jpg" /> are the leading terms of <img src="6-4900022\4b361f5e-fbc1-4a3b-893e-4d24a5377a6b.jpg" /> and <img src="6-4900022\6850231b-0971-4391-9a1e-950a29b6dfa5.jpg" /> respectively.</p><p>If the plate is laminated, it means that the plate is not homogeneous. Equations (2.3) and (2.9) are not equivalent: the mean value is different from the value on each point of the lateral surface. Consequently, the asymptotic solution is not valid everywhere on the plate. Close to the edge, a corrective term must be added.</p></sec><sec id="s2_2"><title>2.2 Behavior of the Thin Plate Close to a Classical edge</title><p>It can be important to have a good approximation of the boundary of the plate, if for instance the damage is studied. As a matter of fact the cracks appear on the edge, like delamination. In this case, a local three-dimensional description of the behavior of the plate is necessary.</p><p>On the lateral edge the assumption of very small <img src="6-4900022\182e8395-4773-4cd8-a937-5ce7fb41d2ee.jpg" /> with respect to the other variables is not justified. Distance to the edge is of the same range of magnitude than the thickness of the plate.</p><p>Insert <xref ref-type="fig" rid="fig3">Figure 3</xref> The specific directions.</p><p>In order to study the temperature close to the edge, let us define local axes <img src="6-4900022\679f8d88-3fc9-46e5-ab8c-7241e8dac594.jpg" /> (cf. <xref ref-type="fig" rid="fig3">Figure 3</xref>): <img src="6-4900022\ef24f41e-b908-487a-82a0-584488a63276.jpg" />is the tangent direction of the edge of the middle plane<img src="6-4900022\15c64d88-1808-4599-a889-998966759b8f.jpg" />, <img src="6-4900022\ca317db7-ee7e-499a-936f-509eb583c146.jpg" />is normal to <img src="6-4900022\0b1fbdb5-af09-43f7-b9ba-b585fc4b22f5.jpg" /> in the middle plane, pointing inside<img src="6-4900022\52d94928-eacf-4e14-ba20-763c4cf38255.jpg" />, <img src="6-4900022\a30d1e3f-59d6-48c7-a743-fb558769032e.jpg" />is normal to<img src="6-4900022\87b7d597-11bb-49e3-9a16-f0af6c7bdaab.jpg" />.</p><p>A corrective term is added to the asymptotic expansion of the temperature [<xref ref-type="bibr" rid="scirp.5774-ref1">1</xref>] to improve it. In order to act on the leading term of the thermal flux, we have to correct the second term of the temperature. As a matter of fact, if we act on the first term<img src="6-4900022\28dc6faa-1f87-4cfb-ab10-b4cad5a43d1f.jpg" />, we shall change the order of the thermal flux.</p><p>Let <img src="6-4900022\568a8320-3c79-41bf-9705-3e64b2402a34.jpg" /> be the corrective term of the temperature, the new asymptotic expansion close to the edge is</p><p><img src="6-4900022\a771871d-f084-46a1-8311-b15f7eb2282e.jpg" />(2.10)</p><p>The corrective term must depend on the position in the thickness <img src="6-4900022\94646543-1222-4860-bb09-68dc6a03c9e4.jpg" /> and on the distance to the lateral edge<img src="6-4900022\eb71b03c-b698-41cd-bdb5-bfb3741d1523.jpg" />. The position on the edge <img src="6-4900022\8693d662-6a3a-44dc-bd5b-50a7841fc51a.jpg" /> is a parameter. So that, it is defined in a semi infinite strip,</p><disp-formula id="scirp.5774-formula123022"><label>(2.11)</label><graphic position="anchor" xlink:href="6-4900022\e0f2350d-8f26-4ee9-9665-844a1cd9d7b2.jpg"  xlink:type="simple"/></disp-formula><p>It can be proved [<xref ref-type="bibr" rid="scirp.5774-ref6">6</xref>] that <img src="6-4900022\84d6007a-4296-4867-918b-a1d9031099dd.jpg" /> is the unique solution of the variational problem</p><p><img src="6-4900022\709325f8-279a-4e9f-8dc2-32700ea2bf85.jpg" /></p><p><img src="6-4900022\c8b28ec3-593c-42fb-a021-b4c6145337f6.jpg" /></p><disp-formula id="scirp.5774-formula123023"><label>(2.12)</label><graphic position="anchor" xlink:href="6-4900022\cf2fc96a-b928-48ae-b810-bb5bf4949d3b.jpg"  xlink:type="simple"/></disp-formula><p>where <img src="6-4900022\11e70645-0497-4ec8-9cd3-ff97534703f3.jpg" /> is the completed space of <img src="6-4900022\15a43075-1597-46fd-b276-2af51ba6de6f.jpg" /> for the norm associated with</p><disp-formula id="scirp.5774-formula123024"><label>(2.13)</label><graphic position="anchor" xlink:href="6-4900022\76361457-e3bd-4c56-a5fb-ab50eda374d1.jpg"  xlink:type="simple"/></disp-formula><p>The separation of variables method [6,7], allows to assume that the corrective term decreases exponentially.</p></sec></sec><sec id="s3"><title>3. Boundary Layer Close to an Angle</title><p>As it was seen in subsection 2.2, the corrective term allows to improve the description of the thermal flux close to a regular edge of the plate. In the same way, the behavior of the plate close to an angle, can be given by an asymptotic expansion with a new corrective term, a new boundary layer term.</p><p>Now, the boundary of the plate is not regular, it admits a perturbation. Close to the origin, the edge forms an angle of magnitude <img src="6-4900022\ce3492bc-a031-4adc-be4d-5ce0cfb87bfd.jpg" /> (cf.<xref ref-type="fig" rid="fig4">Figure 4</xref>). The middle plane of the plate is no more regular. The lateral edge <img src="6-4900022\764c190a-aa96-41f5-bd85-35f51c590bc3.jpg" /> can be split into two parts: <img src="6-4900022\5e664273-9a42-4cd0-a8a9-480a465edf59.jpg" />and<img src="6-4900022\535bff30-ffcd-4a04-8f58-adc37bb88b7e.jpg" />.</p><p><xref ref-type="fig" rid="fig4">Figure 4</xref> The perturbation of the edge In order to lighten the notations, we shall assume that <img src="6-4900022\d0b97cf2-2cae-40aa-a4a9-c59c48bd9d08.jpg" /> vanishes in the vicinity of the origin.</p><p>The edge is assumed to be regular everywhere but not in the vicinity of the origin.</p><p>Far from the origin, but on the lateral edge, the boundary layer problems are similar to those described in section 2, corresponding to a classical, regular surface.</p><p>Let <img src="6-4900022\ebf17e1b-91c9-486d-8f79-b2910704db09.jpg" /> denotes the corrective term on <img src="6-4900022\69b26aee-7333-4ca7-9df9-8eb82d2265ba.jpg" /> far from the origin. In the same way, let <img src="6-4900022\a22844b8-0077-45a2-afcd-aecfedaa4c1a.jpg" /> denotes the corrective term on <img src="6-4900022\23a52b95-7e96-4e50-9232-01e2e98d740f.jpg" /> far from the origin. These functions<img src="6-4900022\0cab78a1-cc14-4906-b8c9-a0a0b14a96f7.jpg" />, <img src="6-4900022\39a7628a-60fa-4472-a022-af0b2733672c.jpg" />, are solutions of the variational problem (2.12) with unknown <img src="6-4900022\ea3f4423-e9c3-4260-8f61-f369177b86ea.jpg" /> instead of<img src="6-4900022\78e27a3c-a6ef-4395-886d-c398d57f0c94.jpg" />.</p><p>We can also prove that they are solutions of</p><disp-formula id="scirp.5774-formula123025"><label>(3.1)</label><graphic position="anchor" xlink:href="6-4900022\e181367d-72f1-4160-95a5-cf52218e7a6a.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123026"><label>(3.2)</label><graphic position="anchor" xlink:href="6-4900022\5d68eaa9-184f-44e2-a50e-94247918e40b.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123027"><label>(3.3)</label><graphic position="anchor" xlink:href="6-4900022\d99e0b93-7691-467a-8e2b-fb4b5afee2b2.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123028"><label>(3.4)</label><graphic position="anchor" xlink:href="6-4900022\e49c9452-1c56-44eb-b342-f9a6d2dbc535.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123029"><label>(3.5)</label><graphic position="anchor" xlink:href="6-4900022\11588491-7753-4da2-9033-eb39e3e26b48.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123030"><label>(3.6)</label><graphic position="anchor" xlink:href="6-4900022\97150258-09b1-4668-a076-3d61b0b50b09.jpg"  xlink:type="simple"/></disp-formula><p>where <img src="6-4900022\da2e176c-1dfc-4a53-9e63-8ea7b2570824.jpg" /> is defined in (2.11).</p><p>Close to the origin, a corrective term, <img src="6-4900022\bd843190-1112-4cd0-b79b-4974cd126877.jpg" />is added to <img src="6-4900022\c6c73b1a-043a-436d-aa9f-ef490cb48427.jpg" /> in the asymptotic expansion (2.7). It depends on the three space components. It is defined in an unbounded domain <img src="6-4900022\5ec7c297-a32e-481f-a0a2-3fd6f5aaf24e.jpg" /> which is the dilatation of the origin. As a matter of fact, close to the origin, all directions (the position in the thickness and the distance to the origin) have the same range of order. Because of the geometry of the domain, it will be useful to introduce the cylindric coordinates to describe the domain:</p><p><img src="6-4900022\0009b292-6d9d-4298-8a95-93f487142258.jpg" />.</p><p>The asymptotic expansion of the temperature is now</p><p><img src="6-4900022\3f0cc757-8435-4f29-bc0c-76c7e319d7c3.jpg" /></p><p>where the unknown is<img src="6-4900022\5af9462f-f035-4a8b-8296-e79b629d8d62.jpg" />.</p><p>When<img src="6-4900022\96492960-bed8-49ff-846b-cb2cc0808976.jpg" />, the boundary condition (2.3) must be exactly satisfied at the corresponding order. When <img src="6-4900022\5c56435b-0c9c-4807-a144-7be189484b01.jpg" /> becomes great, the corrective term must tend to the classical boundary layer term. We shall gather <img src="6-4900022\f37e8b03-0066-41f4-a7b3-53e16d18a5dc.jpg" /> and <img src="6-4900022\62286930-9ecd-4556-aa84-f207d3a5d541.jpg" /> into a unique function <img src="6-4900022\63b8078c-f634-4d94-a15e-ce63c6a1a3b0.jpg" /> defined by</p><p><img src="6-4900022\27ef726c-253c-4097-89c3-4a60b58b6e4c.jpg" /></p><p>Remark 2. For small values of <img src="6-4900022\65101ae7-1654-4e91-985e-418d77bab0e2.jpg" /> and great values of<img src="6-4900022\27ac05c0-3389-493e-91e9-7d844859c195.jpg" />, it means far from the lateral edge, the influence of each corrective term <img src="6-4900022\6a27996e-7eb9-442f-a8b1-3d146378f92a.jpg" /> is very small because of the exponentially decreasing. So that they can be neglected.</p><p>We can then see that, because <img src="6-4900022\ccc2b89f-6a90-40a3-a80e-e19b79f5388f.jpg" /> is unbounded, and because the corrective term must tend to <img src="6-4900022\3655861c-f8b5-4908-9903-21c0be7a2328.jpg" /> when <img src="6-4900022\bc4670ca-b190-4747-a85d-bb26351f7644.jpg" /> becomes great, it cannot belong to <img src="6-4900022\f2e1503f-7aeb-4d21-9610-ff0dea9cbbc9.jpg" />We shall transform the corrective term in order to obtain an unknown which belongs to<img src="6-4900022\f498e535-0915-407c-89c1-f47e6118d952.jpg" />. Let us define:</p><p><img src="6-4900022\f85dee9a-8712-4350-9be7-d6422f9d33fa.jpg" /></p><p>where the new unknown is now<img src="6-4900022\dde350b6-d8a0-4de2-8c6c-cf87dfc1e31a.jpg" />.</p><p>The function<img src="6-4900022\65827dd8-64fc-49ab-abe5-ad7a3caa7d05.jpg" />, which is<img src="6-4900022\6d540095-c3cf-4e09-a775-1c83b81f7266.jpg" />, is a cut off function such that for little <img src="6-4900022\ab62c4c0-5893-4fd3-9a24-9240b5a65280.jpg" /> is equal to 0 and for great value of <img src="6-4900022\5d5d6ec7-554e-4649-94e4-0d1dc6ffc2bd.jpg" /> is equal to 1 (cf. <xref ref-type="fig" rid="fig5">Figure 5</xref>).</p><p>insert figure 5 The cutoff function <img src="6-4900022\3fb451e6-6d09-4ac1-83d8-abd9d726bf3e.jpg" /></p><p>The problem for <img src="6-4900022\5809a174-8c65-42e0-bc6d-d154ae61dc5b.jpg" /> is now:</p><disp-formula id="scirp.5774-formula123031"><label>(3.7)</label><graphic position="anchor" xlink:href="6-4900022\83bb3802-d290-4d71-93c5-8a8fc2195fe1.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123032"><label>(3.8)</label><graphic position="anchor" xlink:href="6-4900022\54d0eaa6-9dd0-4e7b-bdd0-1f0b5638ce38.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123033"><label>(3.9)</label><graphic position="anchor" xlink:href="6-4900022\bf1efd50-9071-42d8-ba56-225ec010a21c.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123034"><label>(3.10)</label><graphic position="anchor" xlink:href="6-4900022\b3805563-764d-4c3f-ac90-50dfefb9f266.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123035"><label>(3.11)</label><graphic position="anchor" xlink:href="6-4900022\e9cec3c0-0f67-4332-bed2-a525e5ee0e11.jpg"  xlink:type="simple"/></disp-formula><disp-formula id="scirp.5774-formula123036"><label>(3.12)</label><graphic position="anchor" xlink:href="6-4900022\690861bc-c655-4aa4-b050-db79d80c5ecc.jpg"  xlink:type="simple"/></disp-formula><p>Equation (3.7) is the equilibrium equation, (3.8) is the boundary condition on <img src="6-4900022\54929d1a-4c44-45bc-864d-e958277fcb7e.jpg" /> and<img src="6-4900022\b3574ef6-eed3-47c1-92b1-ffee65e5a732.jpg" />, (3.11) and (3.12) are the continuity conditions across the interfaces, (3.9) is the boundary condition on the lateral edge, and (3.10) means that the corrective term is inefficient far from the origin.</p><p>Remark 3. When <img src="6-4900022\02d090b1-967f-4929-8d07-f97cb526ff43.jpg" /> is sufficiently great, <img src="6-4900022\090de81b-36f9-4a8d-91d4-814f92ec8d82.jpg" />is equal to 1 and the right-hand sides of (3.7), (3.8), (3.9) and (3.12) vanish.</p><p>Problem (3.7)-(3.12) is equivalent to the following variational problem:</p><p><img src="6-4900022\4fff14d7-2b8e-4d21-a8ba-061b473fcfa3.jpg" /></p><disp-formula id="scirp.5774-formula123037"><label>(3.13)</label><graphic position="anchor" xlink:href="6-4900022\19676927-5b2a-41be-8999-dbbc8f85100b.jpg"  xlink:type="simple"/></disp-formula><p>With</p><p><img src="6-4900022\7c4d1ba8-6b14-4f69-9623-ffc83753cd0b.jpg" /></p><p><img src="6-4900022\23e34b9f-ece5-4c7a-85e5-e7a414af8d09.jpg" /></p><p><img src="6-4900022\0224145e-ff00-4fe9-a5a2-363b603afeda.jpg" /></p><p>Where <img src="6-4900022\7e03585b-1e94-41e5-9ba3-b24b56deb547.jpg" /> is the completed space of <img src="6-4900022\b4f6e332-b275-40d7-bb6b-e0bdff874f78.jpg" /> for the norm associated with</p><p><img src="6-4900022\dc82f001-9a12-44bc-9ed9-e69f522b5ee8.jpg" /></p><p>Lemma 1. The right-hand side of (3.13) is a functional over<img src="6-4900022\cfb31b41-ffcf-4380-bd61-2c2e4cd59d33.jpg" />.</p><p>Proof. <img src="6-4900022\44b9ea19-0ba9-4ab9-9b32-29604a926014.jpg" />is defined over a space of equivalent classes. Two elements of a same class differ by a constant. It follows that <img src="6-4900022\f4e52774-1f1e-4e2f-aef7-8e0c3f414ae0.jpg" /> is a functional over <img src="6-4900022\7d89f8c7-0cb6-49da-aac1-553722dd7284.jpg" /> if two elements of a same class take the same value by<img src="6-4900022\13bfe745-393c-42f0-a486-8e589c1d129c.jpg" />, or if, for any constant<img src="6-4900022\8c3eb4a7-f5d0-4b92-ac79-50f2b2de2af1.jpg" />,<img src="6-4900022\d729be2c-2b16-404b-b5ce-529ba325d2f3.jpg" />. Using the first expression of <img src="6-4900022\1a1ffab9-029f-499c-9d01-d729b270bc07.jpg" /> in (3.14), we get</p><p><img src="6-4900022\5004389d-7b6a-48be-a181-66889d2121cd.jpg" /></p><p>by virtue of (2.9) and the assumption on<img src="6-4900022\52962726-3af1-46f1-99f2-a11ee0e7d301.jpg" />.</p><p>Lemma 2. The functional <img src="6-4900022\38a87c8b-e1d8-4d9a-8da1-f9ed532d4290.jpg" /> is bounded over<img src="6-4900022\06ac8be3-f2cd-4e2f-bc05-1bf26dcfadc7.jpg" />. It means that there exists a constant <img src="6-4900022\036ea2bc-2c2b-4f18-913f-1889612bb68e.jpg" /> such that for all<img src="6-4900022\bd81074b-7b99-48b0-9504-aa5f8d54a9c8.jpg" />,<img src="6-4900022\9aa7d2f6-01bc-447d-9539-a047bb5be48e.jpg" />.</p><p>Proof. Let <img src="6-4900022\83126fd9-aa39-4faa-8e69-69ff2fdbf7a3.jpg" /> be any element of<img src="6-4900022\cfc7624b-cf3c-4086-9570-3287633e1e02.jpg" />, using the second expression of <img src="6-4900022\925ecc0f-a637-43be-886c-25fb7d317328.jpg" /> in (3.14)</p><p><img src="6-4900022\1fcbfe26-b766-48eb-a28f-06391aa54913.jpg" /></p><p>By virtue of remark 3 each integral can be applied on a bounded domain which does not depend on<img src="6-4900022\31871804-cff9-43b1-9e1e-916474e1d682.jpg" />. As a consequence, <img src="6-4900022\1fed7a1e-51ef-47e5-81aa-d076728e4547.jpg" />can be read</p><p><img src="6-4900022\c50af87c-21da-49ba-9433-b7a27e774a8c.jpg" /></p><p>where the upper-script <img src="6-4900022\74d1666b-edd5-44c9-bc9b-f428c7f7cb46.jpg" /> means that the domain is bounded. Using the Cauchy-Schwarz inequality,</p><p>&#160; <img src="6-4900022\c54c658d-76f9-4efd-9ff6-f82d072193eb.jpg" /></p><p>Using the trace theorem and because <img src="6-4900022\e1338f8e-5ab2-4daa-9306-6c8e45c75cab.jpg" /> is a bounded domain, we get</p><p><img src="6-4900022\6407aed5-2a78-4fc0-8168-6ed111b33e47.jpg" /></p><p>Passing to the quotient space <img src="6-4900022\740e1dfa-b4e8-42a0-a2ab-ce1b697978b7.jpg" /></p><p><img src="6-4900022\2d751a61-0a82-4023-9e5b-4428975c7950.jpg" />but</p><p><img src="6-4900022\18ccf22e-6ddd-4e94-b9f4-62e636c748e0.jpg" /></p><p>Because of the density of <img src="6-4900022\9817cf87-24d8-453b-9005-f5f425b6c62e.jpg" /> into<img src="6-4900022\c159ddab-de7a-4a60-8af6-ef04f0bdeabf.jpg" />, Equation (3.13) is valid in the whole<img src="6-4900022\aef1b2be-50f9-4cc9-b622-f917b8519c5c.jpg" />. It follows from the Lax-Milgram theorem, lemma 2 that Theorem. The corrective term <img src="6-4900022\d94bb74d-9a6f-4e0e-a38a-084e2ced89a5.jpg" /> is uniquely defined over<img src="6-4900022\16ed1c3d-bd97-4606-ac96-55303a4c8dc1.jpg" />.</p></sec><sec id="s4"><title>4. Conclusions</title><p>In order to improve the description of the behavior of the plate close to the singularity, a boundary layer term was added. This term is solution of (3.7)-(2.12). It has no influence far from the edge but it is defined over an unbounded domain.</p><p>At first, the equivalent variational problem was found. Then, the previous theorem allows us to prove the existence and the uniqueness of the solution. In this way the numerical resolution can be implement.</p></sec><sec id="s5"><title>REFERENCES</title></sec></body><back><ref-list><title>References</title><ref id="scirp.5774-ref1"><label>1</label><mixed-citation publication-type="other" xlink:type="simple">H. Dumontet, “Homogénéisation et Effets de Bords dans les Matériaux Composites,” Ph. D. Thesis, Paris, 1990.</mixed-citation></ref><ref id="scirp.5774-ref2"><label>2</label><mixed-citation publication-type="other" xlink:type="simple">I. Titeux, “Problèmes de Fracture et Comportement Local Pour les Plaques Composites Anisotropes,” Ph. D. Thesis, Paris, 1995.</mixed-citation></ref><ref id="scirp.5774-ref3"><label>3</label><mixed-citation publication-type="other" xlink:type="simple">A. R. Saidi, F. Hejripour and E. Jomehzadeh, “On the Stress Singularities and Boundary Layer in Moderately Thick Functionally Graded Sectorial Plates,” Applied Mathematical Modelling, Vol. 34, No. 11, 2010, pp 3478-3492. doi:10.1016/j.apm.2010.02.036</mixed-citation></ref><ref id="scirp.5774-ref4"><label>4</label><mixed-citation publication-type="other" xlink:type="simple">E. Jomehzadeh and A. R. Saidi, “Analytical Solution for Free Vibration of Transversely Isotropic Sector Plates Using a Boundary Layer Function,” Thin Walled Structures, Vol. 47, No. 1, 2009, pp. 82-88. 
doi:10.1016/j.tws.2008.05.004</mixed-citation></ref><ref id="scirp.5774-ref5"><label>5</label><mixed-citation publication-type="other" xlink:type="simple">J. Sanchez-Hubert and E. Sanchez-Palencia, “Intro- duction Aux Methods Asymptotiques et à L'homo- généisation,” Masson, Paris, 1992.</mixed-citation></ref><ref id="scirp.5774-ref6"><label>6</label><mixed-citation publication-type="other" xlink:type="simple">D. Leguillon and E. Sanchez-Palencia, “Computation of Singular Solutions in Elliptic Problems and Elasticity,” Masson, Paris, 1988.</mixed-citation></ref><ref id="scirp.5774-ref7"><label>7</label><mixed-citation publication-type="other" xlink:type="simple">I. Titeux and Y. Yakubov, M3AS Vol. 7, No. 7, 1997, pp. 1035-1050.</mixed-citation></ref></ref-list></back></article>