A COMPARISON OF THE CONTINUOUS AND DISCRETE ADJOINT APPROACH TO AUTOMATIC AERODYNAMIC OPTIMIZATION
|
|
- Neil Garrison
- 5 years ago
- Views:
Transcription
1 AIAA--667 A COMPARISON OF THE CONTINUOUS AND DISCRETE ADJOINT APPROACH TO AUTOMATIC AERODYNAMIC OPTIMIZATION Siva K. Nadarajah and Antny Jamesn Department f Aernautics and Astrnautics Stanfrd University Stanfrd, Califrnia 9435 U.S.A. Abstract This paper cmpares the cntinuus and discrete adjint-based autmatic aerdynamic ptimizatin. The bjective is t study the trade-ff between the cmplexity f the discretizatin f the adjint equatin fr bth the cntinuus and discrete apprach, the accuracy f the resulting estimate f the gradient, and its impact n the cmputatinal cst t apprach an ptimum slutin. First, this paper presents cmplete frmulatins and discretizatin f the Euler equatins, the cntinuus adjint equatin and its cunterpart the discrete adjint equatin. The differences between the cntinuus and discrete bundary cnditins are als explred. Secnd, the results demnstrate tw-dimensinal inverse pressure design and drag minimizatin prblems as well as the accuracy f the sensitivity derivatives btained frm cntinuus and discrete adjint-based equatins cmpared t finite-difference gradients. Intrductin In the 97s several attempts were made t use Cmputatinal Fluid Dynamics (CFD as a design tl. 3 Since then CFD has had a significant impact. Many individuals have refcused their attentin n autmatic aerdynamic ptimizatin, because f accurate numerical schemes and an expnential grwth in cmputatinal speed at affrdable prices. The mathematical thery fr cntrl systems gverned by partial differential equatins has created a framewrk fr the frmulatin f inverse design and general aerdynamic prblems at a reduced cmputatinal cst. 4, 5 Recently, with the help f Graduate Student, Student Member AIAA Thmas V. Jnes Prfessr f Engineering, Stanfrd University, AIAA Fellw Cpyright c 999 by Siva Nadarajah and Antny Jamesn a new generatin f cmputers, autmatic aerdynamic ptimizatin has been revisited. 6 Optimizatin techniques fr the design f aerspace vehicles generally use gradient-based methds in which the vehicle shape is parameterized with a set f design variables. A feasible ptimum shape is nly achievable with an apprpriate cst functin. Typically such cst functins are drag cefficients, lift t drag ratis, target pressure distributins, etc. Sensitivity derivatives f the cst functin with respect t the design variables are calculated by taking small steps in each and every design variable. These sensitivity derivatives are then used t get a directin f imprvement and a step is taken until cnvergence is achieved. Each step requires a cmplete flw slutin, and fr a large number f design variables such methds are cmputatinally cstly. The mathematical thery fr the cntrl systems gverned by partial differential equatins, as develped, fr example, by J.L. Lins, 4 decreases the cst and is mre advantageus than the classical finite-difference methds. In cntrl thery the gradient is calculated indirectly by slving the adjint equatin. The cst f btaining the sensitivity derivatives f the cst functin with respect t each design variable frm the slutin f the adjint equatin is negligible in cmparisn with the cst f the flw calculatin. Cnsequently, the ttal cst t btain these gradients is essentially independent f the number f design variables, amunting t the cst f ne flw slutin and ne adjint slutin, where the adjint equatin is a linear equatin and thus f reduced cmplexity. This methd was first applied t transnic flw by Jamesn. 6 In the last six years autmatic aerdynamic design f cmplete aircraft cnfiguratins has been successful, yielding ptimized slutins f wing 4, 6 and wing-bdy cnfiguratins. The cntinuus adjint apprach thery was develped by cmbining the variatin f the cst func- American Institute f Aernautics and Astrnautics
2 tin and field equatins with respect t the flw-field variables and design variables thrugh the use f Lagrange multipliers als called cstate r adjint variables. Cllecting the terms assciated with the variatin f the flw-field variables prduces the adjint equatin and its bundary cnditin. The terms assciated with the variatin f the design variable prduce the gradient. The field equatins and the adjint equatin with its bundary cnditin must be discretized t btain numerical slutins. As the mesh is refined, the cntinuus adjint yields the exact gradient. The discrete adjint apprach means applying the cntrl thery directly t the set f discrete field equatins. The discrete adjint equatin is derived by cllecting tgether all the terms multiplied by the variatin δw f the discrete flw variable. If the discrete adjint equatin is slved exactly, then the resulting slutin fr the Lagrange multiplier prduces an exact gradient f the inexact cst functin and the derivatives are cnsistent with finite difference gradients independent f the mesh size. A subject f n-ging research is the trade-ff between the cmplexity f the adjint discretizatin, the accuracy f the resulting estimate f the gradient, and its impact n the cmputatinal cst t apprach an ptimum slutin. Shubin and Frank presented a cmparisn between the cntinuus and discrete adjint fr quasi-ne-dimensinal flw. A variatin f the discrete field equatins prved t be cmplex fr higher rder schemes. Due t this limitatin f the discrete adjint apprach, early implementatin f the discretizatin f the adjint equatin was nly cnsistent with a first rder accurate flw equatin. Beux and Dervieux 5 used a first rder upwind scheme with Van Leer flux vectr splitting n a tw-dimensinal unstructured grid. Burgreen and Baysal 6 carried a secnd rder implementatin f the discrete adjint n threedimensinal shape ptimizatin f wings fr structured grids. Fr secnd rder accuracy n unstructured grids, Ellit and Peraire 7 slved the Euler equatins by a multistage Runge-Kutta scheme with Re decmpsitin fr the dissipative fluxes n tw and three-dimensinal unstructured grids. They perfrmed ptimizatin n inverse pressure designs f multielement airfils and wing-bdy cnfiguratins in transnic flw. Andersn and Venkatakrishnan 8 cmputed inviscid and viscus ptimizatin n unstructured grids using bth the cntinuus and discrete adjint. Ill, Salas, and Ta saan 9 investigated shape ptimizatin n ne and twdimensinal flws using the cntinuus adjint apprach. Ta saan, Kuruvila, and Salas used a nesht apprach with the cntinuus adjint frmulatins. Kim, Alns, and Jamesn cnducted an extensive gradient accuracy study f the Euler and Navier-Stkes equatins which cncluded that gradients frm the cntinuus adjint methd were in clse agreement with thse cmputed by finite difference methds, and less dependent n the level f cnvergence f the flw slver. Objectives The bjectives f this wrk are:. Review the frmulatin and develpment f the cmpressible adjint equatins fr bth the cntinuus and discrete apprach.. Investigate the differences in the implementatin f bundary cnditins fr each methd. 3. Cmpare the gradients f the tw methds t finite difference gradients fr inverse pressure design and drag minimizatin. 4. Cmpare the cnvergence between the cntinuus and discrete adjint. 5. Study the differences in calculating the exact gradient f the inexact cst functin (discrete adjint r the inexact gradient f the exact cst functin (cntinuus. The Design Prblem as a Cntrl Prblem A simple apprach t ptimizatin is t represent the gemetry thrugh a set f design parameters, which may, fr example, be the weights α i applied t a set f shape functins b i (x s that the shape is represented as f(x = α m b m (x. I α m Next, a cst functin I which is a functin f the weight parameters α m is chsen. Such a cst functin can be the difference between the current and target pressure distributin fr inverse design prblems, drag cefficient fr drag minimizatin prblems, r lift t drag rati. The sensitivities may nw be estimated by making a small variatin δα m in each design parameter in turn and recalculating the flw t btain the change in I. Then, using a finite difference frmula, I I(α m δα m I(α m. α m δα m American Institute f Aernautics and Astrnautics
3 The gradient vectr I α may nw be used t determine a directin f imprvement. The simplest prcedure uses the methd f steepest descent and takes a step in the negative gradient directin by setting α n = α n λ I α, s that t first rder I δi = I IT δα = I λ IT α α I α. The main disadvantageus f the finite difference methd are first that N flw calculatins are needed t calculate the sensitivities f N design variables, and secnd that the accuracy is sensitive t the step size δα m. These difficulties are circumvented by the cntrl thery apprach which may be utlined in abstract frm as fllws. Fr flw arund an airfil, the aerdynamic prperties that define the cst functin are functins f the flw-field variables, w, andthephysical lcatin f the bundary, which may be represented by the functin F, say.then I = I (w, F, and a change in F results in a change δi = IT IT δw δf, ( F in the cst functin. Using cntrl thery the gverning equatins f the flw-field are nw intrduced as a cnstraint in such a way that the final expressin fr the gradient des nt require reevaluatin f the flw-field. In rder t achieve this δw must be eliminated frm (. Suppse that the gverning equatin R which expresses the dependence f w and F within the flw-field dmain D can be written as R (w, F =. ( Then δw is determined frm the equatin [ ] [ ] R R δr = δw δf =. (3 F Next, intrducing a Lagrange Multiplier ψ, wehave ([ ] [ ] I T IT R R δi = δw F δf ψt δw δf F { [ ]} { [ ]} I T R I T R = ψt δw F ψt δf. F Chsing ψ t satisfy the adjint equatin [ ] T R ψ = I (4 the first term is eliminated, and we find that where δi = GδF, (5 G = IT F ψt [ ] R. F Euler Equatins In rder t allw fr gemetric shape changes it is cnvenient t use a bdy fitted crdinate system, s that the cmputatinal dmain is fixed. This requires the frmulatin f the Euler equatins in the transfrmed crdinate system. The Cartesian crdinates and velcity cmpnents are dented by x, x,andu, u. Einstein ntatin simplifies the presentatin f the equatins, where summatin ver k = t is implied by a repeated index k. Then the tw-dimensinal cmpressible Euler equatins may be written as t f k = ind, (6 x k where w = ρ ρu ρu ρe, f k = ρu k ρu k u pδ k ρu k u pδ k ρu k H (7 and δ kl is the Krnecker delta functin. Als, { p =(γ ρ E ( } u k, (8 and ρh = ρe p (9 where γ is the rati f the specific heats. Cnsider a transfrmatin t crdinates ξ, ξ, where [ xk K kl = ξ l and ], J =det(k, K kl = S = JK. [ ξk x l ], The elements f S are the cefficients f K, andin a finite vlume discretizatin they are just the face 3 American Institute f Aernautics and Astrnautics
4 areas f the cmputatinal cells prjected in the x and x directins. Als intrduce scaled cntravariant velcity cmpnents as U k = S kl u l. The Euler equatins can nw be written as where and W t F k ξ k = ind, ( F k = S kl f l = W = Jw, ρu k ρu k u S l p ρu k u S l p ρu k H. ( Assume nw that the new cmputatinal crdinate system cnfrms t the airfil in such a way that the airfil surface B W is represented by ξ =. Then the flw is determined as the steady state slutin f equatin ( subject t the flw tangency cnditin U = nb W. ( At the far field bundary B F, cnditins are specified fr incming waves, as in the tw-dimensinal case, while utging waves are determined by the slutin. When equatin ( is frmulated fr each cmputatinal cell, a system f first-rder rdinary differential equatins is btained. T eliminate dd-even decupling f the slutin and vershts befre and after shck waves, the cnservative flux is added t a diffusin flux. The artificial dissipatin scheme used in this research is a blended first and third rder flux, first intrduced by Jamesn, Schmidt, and Turkel. The artificial dissipatin scheme is defined as, D i,j = ɛ i,j(w i,j w ɛ 4 i,j(w 3w i,j 3w w i,j. (3 The first term in equatin (3 is a first rder scalar diffusin term, where ɛ is scaled by the i,j nrmalized secnd difference f the pressure and serves t damp scillatins arund shck waves. ɛ 4 is the cefficient fr the third derivative f the i,j artificial dissipatin flux. The cefficient is scaled such that it is zer at regins f large gradients, such as shck waves and eliminates dd-even decupling elsewhere. Design using the Euler Equatins This sectin illustrates applicatin f cntrl thery t aerdynamic design prblems fr the case f tw-dimensinal airfil design using the cmpressible Euler equatins as the mathematical mdel. Cntinuus Adjint The weak frm f the Euler equatins fr steady flw is φ T F k dd = n k φ T F k db, (4 D ξ k B where the test vectr φ is an arbitrary differentiable functin and n k is the utward nrmal at the bundary. If a differentiable slutin w is btained t this equatin, then it can be integrated by parts t give φ T F k dd =. ξ k D Since this is true fr any φ the differential frm can be recvered. If the slutin is discntinuus, then (4 may be integrated by parts separately n either side f the discntinuity t recver the shck jump cnditins. Suppse nw that we desire t cntrl the surface pressure by varying the wing shape. Fr this purpse, it is cnvenient t retain a fixed cmputatinal dmain. Variatins in the shape then result in crrespnding variatins in the mapping derivatives defined by K. Intrduce the cst functin I = (p p d ds, B W where p d is the desired pressure. The design prblem is nw treated as a cntrl prblem where the cntrl functin is the wing shape, which is chsen t minimize I subject t the cnstraints defined by the flw equatins (-. A variatin in the shape causes a variatin δp in the pressure and cnsequently a variatin in the cst functin δi = (p p d δp ds (p p d δds. B W B W (5 Since p depends n w thrugh the equatin f state (8 9, the variatin δp is determined frm the variatin δw. Define the Jacbian matrices A k = f k, C k = S kl A l. (6 4 American Institute f Aernautics and Astrnautics
5 The weak frm f the equatin fr δw in the steady state becmes φ T δf k dd = (n k φ T δf k db, ξ k where D B δf k = C k δw δs kl f l, which shuld hld fr any differentiable test functin φ. This equatin may be added t the variatin in the cst functin, which may nw be written as δi = (p p d δp ds (p p d δds B W B W ψ T δf k dd (n k ψ T δf k db (7 ξ k D On the wing surface B W, n =. Thus, it fllws frm equatin ( that S δf = δp δs p. (8 S δp δs p Since the weak equatin fr δw shuld hld fr an arbitrary chice f the test vectr φ, wearefree t chse φ t simplify the resulting expressins. Therefre we set φ = ψ, where the cstate vectr ψ is the slutin f the adjint equatin ψ t CT k B ψ ξ k = ind. (9 At the uter bundary incming characteristics fr ψ crrespnd t utging characteristics fr δw. Cnsequently we can chse bundary cnditins fr ψ such that n k ψ T C k δw =. If the crdinate transfrmatin is such that δs is negligible in the far field, then the nly remaining bundary term is ψ T δf dξ. B W Thus, by letting ψ satisfy the bundary cnditin, ψ j n j = p p d n B W, ( where n j are the cmpnents f the surface nrmal, n j = S j Sj S j we find finally that δi = (p p d δds B W ψ T δs kl f l dd D ξ k (δs ψ δs ψ 3 pdξ. ( B W Numerical Discretizatin The cntinuus adjint equatin is linear and cnsequently it culd be slved by direct numerical inversin. The cst f the assciated matrix inversin can becme prhibitive as the number f mesh cells are increased. Instead, since the equatins are similar t that f the Euler equatins, the same iterative methd is used t slve the cntinuus adjint equatin. In this research, a five stage Runge-Kutta scheme with three evaluatins f the artificial dissipatin scheme is used. We emply the blended first and third rder scalar diffusin scheme used fr the Euler equatins here as well. The fllwing is a secnd rder discretizatin f the cntinuus adjint equatin, ( V ψ [ ] T [ ] T f = y η x η t ( [ ] T [ ] T f y η x η ( [ ] T [ ] T f y ξ x ξ ( [ ] T [ ] T f y ξ x ξ d i,j d i,j d d ( ψi,j ( ψi,j ( ψ ( ψ ( where, V is the cell area and d i,j has the same frm as equatin (3. In the case f the cntinuus adjint bundary cnditin, equatin ( dictates values fr the nrmal adjint velcities. The chice fr ψ, ψ 4,and the tangential adjint velcity are arbitrary, therefre assigning a zer value fr these variables des 5 American Institute f Aernautics and Astrnautics
6 nt vilate equatin (. This results, hwever, in a pr cnvergence fr the adjint equatin since it is an ver-specificatin f the adjint bundary cnditin. A satisfactry bundary cnditin may be frmulated as fllws: ψ i, = ψ i, ( ψ i, = ψ i, n (p pd n ψ i, n ψ 3i, ( ψ 3i, = ψ 3i, n (p pd n ψ i, n ψ 3i, ψ 4i, = ψ 4i, (3 Discrete Adjint The discrete adjint equatin is btained by applying the cntrl thery directly t the set f discrete field equatins. The resulting equatin depends n the type f scheme used t slve the flw equatins. This paper uses, a cell centered multigrid scheme with upwind biased blended first and third rder fluxes as the artificial dissipatin scheme. A full discretizatin f the equatin wuld invlve discretizing every term that is a functin f the state vectr. where, n i = S i Sj S j nx ny ( δi = δi c ψ T δ R (w D (w i= j= (5 The subscript i, andi, in the abve equatins dente cells belw and abve the wall. Here, the first and furth cstate variables belw the wall are set equal t the crrespnding values abve the wall and the tangential adjint velcities abve and belw the wall are equated. Drag Minimizatin If the drag is t be minimized, then the cst functin is the drag cefficient, I = C ( d y = C p c B ( W c ξ dξ B W C p x ξ dξ cs α sin α A variatin in the shape causes a variatin p in the pressure and cnsequently a variatin in the cst functin, δi = c c B W C p B W C p ( y x cs α ξ ( y cs α δ ξ ( δ ξ sin α ( x ξ pdξ sin α dξ (4 As in the inverse design case, the first term is a functin f the state vectr, and therefre is incrprated int the bundary cnditin, where the integrand replaces the pressure difference term in equatin (3. The secnd term is added n t the gradient term. where δi c is the discrete cst functin, R(w isthe field equatin, and D(w is the artificial dissipatin term. Terms multiplied by the variatin δw f the discrete flw variables are cllected and the fllwing is the resulting discrete adjint equatin, V ψ = ( t [ ] T [ ] T f y ηi x ηi,j,j ( [ ] T [ ] T f y ηi x ηi,j,j ( [ ] T [ ] T f x ξ y ξ ( [ ] T [ ] T f x ξ y ξ ( [ ] T [ ] T f y ηi x ηi,j,j ( [ ] T [ ] T f y ηi x ηi,j,j ( [ ] T [ ] T f x ξ y ξ ( [ ] T [ ] T f x ξ y ξ ψ i,j ψ i,j ψ ψ ψ ψ ψ ψ δd i,j δd i,j δd δd. (6 6 American Institute f Aernautics and Astrnautics
7 where, δd i,j = ɛ i,j(ψ i,j ψ ɛ 4 i 3,jψ 3ɛ 4 i,j (ψ i,j ψ ɛ 4 i 3,jψ i,j (7 is the discrete adjint artificial dissipatin term and V is the cell area. The dissipatin cefficients ɛ and ɛ 4 are functins f the flw variables, but t reduce cmplexity they are treated as cnstants. The effect f this partial discretizatin f the artificial dissipatin term is explred in the Results sectin. In the case f an inverse design, δi c is the discrete frm f equatin (5. The δw i, term is added t the crrespnding term frm equatin (6, and the metric variatin term is added t the gradient term. In cntrast t the cntinuus adjint, where the bundary cnditin appears as an update t the cstate variables in the cell belw the wall, the discrete bundary cnditin appears as a surce term in the adjint fluxes. At cell i, the adjint equatin is as fllws, V ψ i, = t [ ] A T i, (ψ i, ψ i, A T i, (ψ i, ψ i, [ ] B T i, (ψ 5 i,3 ψ i, Φ (8 where V is the cell area, Φ is the surce term fr inverse design, Φ= ( y ξ ψ i, x ξ ψ 3i, (p p T s i δpi, and, A T i, = y η i, [ ] T [ ] T f x ηi, i, i, All the terms in equatin (8 except fr the surce term are scaled as the square f x. Therefre, as the mesh width is reduced, the terms within parenthesis in the surce term divided by s i must apprach zer as the slutin reaches a steady state. One then recvers the cntinuus adjint bundary cnditin as stated in equatin (. If a first rder artificial dissipatin equatin is used, then equatin (7 wuld reduce t the term assciated with ɛ. In such a case, the discrete adjint equatins are cmpletely independent f the cstate variables in the cells belw the wall. Hwever, if we use the blended first and third rder equatin, then these values are required. As shwn later, a simple zerth rder extraplatin acrss the wall prduces gd results. Replacing the inverse design bundary cnditin in equatin (8 by the discrete frm f equatin (4 results in a discrete adjint equatin fr drag minimizatin. Optimizatin Prcedure The search prcedure used in this wrk is a simple descent methd in which small steps are taken in the negative gradient directin. Let F represent the design variable, and G the gradient. Then an imprvement can be made with a shape change δf = λg, The gradient G can be replaced by a smthed value G in the descent prcess. This ensures that each new shape in the ptimizatin sequence remains smth and acts as a precnditiner which allws the use f much larger steps. T apply smthing in the ξ directin, the smthed gradient G may be calculated frm a discrete apprximatin t G ɛ G = G ξ ξ where ɛ is the smthing parameter. If the mdificatin is applied n the surface ξ = cnstant, then the first rder change in the cst functin is δi = = λ GδFdξ ( G ɛ ξ ( ( = λ G G ɛ ξ <, G ξ Gdξ dξ assuring an imprvement if λ is sufficiently small and psitive. The smthing leads t a large reductin in the number f design iteratins needed fr cnvergence. An assessment f alternative search methds fr a mdel prblem is given by Jamesn and Vassberg. 3 7 American Institute f Aernautics and Astrnautics
8 Results This sectin presents the results f the inviscid inverse design and drag minimizatin cases. Fr each case, we cmpare the cntinuus and discrete gradients, study the adjint slutins frm each methd, and cmpare the cnvergence f the methds. Grid Size Cnt. Disc. Cnt-Disc 96 x 6 3.6e 3.397e e 4 9 x 3.73e 3.74e 3.3e 4 56 x 64.44e 3.49e e 5 Table : L nrm f the Difference Between Adjint and Finite Difference Gradient Inverse Design The target pressure is first btained using the FLO83flw slver fr the NACA 64A4 airfil at a flight cnditin f M =.74 and a lift cefficient f C l =.63n a 9 x 3 C-grid. At such a cnditin the NACA 64A4 prduces a strng shck n the upper surface f the airfil, thus making it an ideal test case fr the adjint versus finite difference cmparisn. The gradient fr the cntinuus and discrete adjint is btained by perturbing each pint n the airfil. We apply an implicit smthing technique t the gradient, befre it is used t btain a directin f descent fr each pint n the surface f the airfil. Figure ( illustrates an inverse design case f a Krn t NACA 64A4 airfil at fixed lift cefficient. Figure (a shws the slutin fr the Krn airfil at M =.74 and C l =.63. After five design cycles we achieve a general shape f the target airfil as shwn in figure (b. After twenty five design cycles the upper surface shape is btained, and nearly eighty percent f the lwer surface is achieved. Fllwing a few mre iteratins, we btain the desired target pressure except fr a few pints at the trailing edge. Observe the pint-t-pint match-up at the shck. Figures (, (3, and (4 exhibit the values f the gradients btained frm the adjint methds and finite difference fr varius grid sizes. The circles dente values that we btain by using the finite difference methd. The square represents the discrete adjint gradient. The asterisk represents the cntinuus adjint gradient. The gradient is btained with respect t variatins in Hicks-Henne sine bump functins placed alng the upper and lwer surface f the airfil. 3, The figures nly illustrate the values btained frm the upper surface starting frm the leading edge n the left and ending at the trailing edge n the right. In rder t reach an accurate finite difference gradient, we btain gradients fr varius step sizes until the finite difference gradient fr each pint cnverges. The discrete adjint equatin is btained frm the discrete flw equatins but withut taking int accunt the dependence f the dissipa- tin cefficients n the flw variables. Therefre, in rder t eliminate the effect f this n cmparisns with the finite difference gradient we cmpute the flw slutin until attaining a decrease f seven rders f magnitude in the residue. We then freeze the dissipative cefficients and calculate the finite difference value fr each design pint. The figures shw that the nly discrepancies exist in the trailing edge area. Table cntains values f the L nrm f the difference between the adjint and finite difference gradients. The table illustrates three imprtant facts: the difference between the cntinuus adjint and finite difference gradient is slightly greater than that between the discrete adjint and finite difference gradient; the nrm decreases as the mesh size is increased; and the difference between cntinuus and discrete adjint gradients decreases as the mesh size is reduced. The secnd clumn depicts the difference between the cntinuus adjint and finite difference gradient. The third clumn depicts the difference between the discrete adjint and finite difference gradients. The last clumn depicts the difference between the discrete adjint and cntinuus adjint. As the mesh size increases the nrms decrease as expected. Since we derive the discrete adjint by taking a variatin f the discrete flw equatins, we expect it t be cnsistent with the finite difference gradients and thus t be clser than the cntinuus adjint t the finite difference gradient. This is cnfirmed by numerical results, but the difference is very small. As the mesh size increases, the difference between the cntinuus and discrete gradients shuld decrease, and this is reflected in the last clumn f table. Figure (5 presents the effect f the partial discretizatin f the flw slver t btain the discrete adjint equatin. Here we btain the finite difference gradients in the figure withut freezing the dissipative cefficients. A small discrepancy exists in regins clser t the leading edge and arund the shck. Kim, Alns, and Jamesn verified that accurate finite difference gradients require a cnvergence 8 American Institute f Aernautics and Astrnautics
9 f fur t five rders f magnitude in the flw slver. Hwever, bth the cntinuus and discrete adjint gradients nly require a cnvergence f tw rders f magnitude in the flw slver. Figures (6 and (7 illustrate the cntinuus and discrete gradients fr varius flw slver cnvergence. In figure (8 and (9 cntinuus and discrete adjint gradients are pltted fr varius adjint slver cnvergence. The gradients nly require tw rders f magnitude cnvergence in the adjint slver. Figure ( shws a cmparisn f the prfiles f the secnd and third cstate values between the cntinuus and discrete adjint methd in a directin nrmal t the bundary. The slutins agree in the interir pints, differing nly at the cell belw the bundary due t the different treatment f the bundary cnditin. In the cntinuus case the value at cell ne is updated by the bundary cnditin. This is in cntrasts t the discrete case where the bundary cnditin appears as a surce term when the fluxes are accumulated in cell tw and the bundary cnditin des nt depend n the value f the cstate in cell ne. In figure ( bth methds prduce similar cnvergence histries. In figure ( we attempt t design a Krn airfil based n the target pressure f the NACA 64A4 at a Mach number f.78. Bth the initial and target pressures cntain a very strng shck. A cmparisn f the finite difference and adjint gradients reveals an increase in the discrepancy between the tw gradients in the vicinity f the shck. In cntrast t figure (, where the shck lcatin is at mesh pint 75 alng the surface, figure (3 illustrates the discrepancy arund the strnger shck arund mesh pint 8. Drag Minimizatin The cst functin fr drag minimizatin is the pressure drag f the airfil. We perfrm cmputatins n a NACA 64A4 airfil at a flight cnditin f M =.75 and fixed lift cefficient f C l =.63. As befre, the gradients were btained by taking variatins respect t Hicks-Henne sine bump functins placed alng the upper and lwer surface f the airfil. Figure (5a illustrates the initial slutin f the airfil with 3 drag cunts. After tw design cycles, the drag is reduced by a third t 44 drag cunts. The strng shck in the initial slutin is weakened. And after just fur design cycles, this value is further halved. In figure (5d, the final design des nt cntain any shck and the drag cunt is a mere 5. Grid Size Cnt. Disc. Cnt-Disc 96 x 6.9e.75e.9e 9 x 3.49e 7.577e 3 5.7e 3 56 x e e 3.35e 3 Table : L nrm f the Difference Between Adjint and Finite Difference Gradient Figures (6-8 illustrate the values f the gradients btained frm the adjint methds and finite difference fr varius grid sizes. The finite difference gradients are based n the same methd used fr the inverse design case, where the dissipative cefficients are frzen after a cnverged flw slutin is btained t simulate a full discretizatin fr the discrete adjint equatin. We reduce the finite difference step sizes until we gain cnverged values fr each design pint. We plt gradients fr the upper surface frm leading edge t trailing edge. In figure (6 design pints between 5 and 6 are lcated in the vicinity f the leading edge, where the gradient has a psitive slpe. In this regin the discrete adjint gradient agrees better with the finite difference gradient, if cmpared t the cntinuus adjint gradient. The difference reduces as the grid size increases. Apart frm the regin f the leading edge, the adjint and finite difference gradients agree. Table ( cntains values f the L nrm f the difference between the adjint and finite difference gradients. Similar t the inverse design case, the table illustrates three imprtant facts: the discrete adjint gradient is clser than the cntinuus adjint gradient t the finite difference gradient; the nrms decrease as the mesh size increases; and, finally, the difference between the cntinuus and discrete adjint gradient decreases as the mesh size increases. We recalculate the finite difference and adjint gradients in figure (9 fr the medium size mesh f 9 x 3 cells t illustrate the effect f partial discretizatin f the flw slver. The dissipative cefficients are nt frzen during the finite difference calculatins. A very small discrepancy appears in the leading edge and in the shck wave (pints: Figures ( and ( illustrate the cntinuus and discrete gradients fr varius flw slver cnvergence. Only a single rder magnitude drp in the flw slver is required fr the adjint gradients t cnverge. We plt cntinuus and discrete adjint gradients in figure ( and (3 fr varius adjint slver cnvergence. The gradients nly require ne rder f magnitude cnvergence in the adjint slver. 9 American Institute f Aernautics and Astrnautics
10 Figure (4 shws a cmparisn f cnvergence f the bjective functin between the cntinuus and discrete adjint. Bth methds cnverge t the same value fr the bjective functin. Figure (5 presents the secnd and third cstate prfiles nrmal t the bundary fr the cntinuus and discrete adjint slutins. Bth slutins agree in the interir pints but disagree at the cell belw the wall. This is due t the difference between the enfrcement f the bundary cnditin. Figure (6 shws that bth adjint methds prduce the same cnvergence histry. Cnclusin This paper presents a cmplete frmulatin fr the cntinuus and discrete adjint apprach t autmatic aerdynamic design using the Euler equatins. The gradients frm each methd are cmpared t finite difference gradients. We cnclude:. The cntinuus bundary cnditin appears as an update t the cstate values belw the wall fr a cell-centered scheme, and the discrete bundary cnditin appears as a surce term in the cell abve the wall. As the mesh width is reduced, ne recvers the cntinuus adjint bundary cnditin frm the discrete adjint bundary cnditin. (Equatins 3and 8. Discrete adjint gradients have better agreement than cntinuus adjint gradients with finite difference gradients as expected, but the difference is generally small. (Figure 6 3. As the mesh size increases, bth the cntinuus adjint gradient and the discrete adjint gradient apprach the finite difference gradient. (Figures The difference between the cntinuus and discrete gradient reduces as the mesh size increases. (Tables and 5. The cst f deriving the discrete adjint is greater. (Equatin 6 6. With ur search prcedure as utlined, the verall cnvergence f the bjective functin is nt significantly affected when the discrete adjint gradient is used instead f the cntinuus adjint gradient. Cnsequently, we find n particular benefit in using the discrete adjint methd, which requires greater cmputatinal cst. Hwever, we believe it beneficial t use the discrete adjint equatin as a guide fr the discretizatin f the cntinuus adjint equatin. (Figure 4 Acknwledgments This research has benefited greatly frm the generus supprt f the AFOSR under grant number AF F The first authr wuld like t thank Juan Alns and James Reuther, fr many useful discussins. References [] F. Bauer, P. Garabedian, D. Krn, and A. Jamesn Supercritical Wing Sectins II. Springer-Verlag, New Yrk, 975 [] P. Garabedian, and D. Krn Numerical Design f Transnic Airfils Prceedings f SYNSPADE 97. pp 53-7, Academic Press, New Yrk, 97. [3] R. M. Hicks and P. A. Henne. Wing Design by Numerical Optimizatin. Jurnal f Aircraft. 5:47-4, 978. [4] J.L. Lins. Optimal Cntrl f Systems Gverned by Partial Differential Equatins. Springer-Verlag, New Yrk, 97. Translated by S.K. Mitter. [5] O. Pirnneau. Optimal Shape Design fr Elliptic Systems. Springer-Verlag, New Yrk, 984. [6] A. Jamesn. Aerdynamic design via cntrl thery. In Jurnal f Scientific Cmputing, 3:33-6,988. [7] A. Jamesn. Autmatic design f transnic airfils t reduce the shck induced pressure drag. In Prceedings f the 3st Israel Annual Cnference n Aviatin and Aernautics, Tel Aviv, pages 5 7, February 99. [8] A. Jamesn. Optimum aerdynamic design using CFD and cntrl thery. AIAA paper 95-79, AIAA th Cmputatinal Fluid Dynamics Cnference, San Dieg, CA, June 995. [9] A. Jamesn., N. Pierce, and L. Martinelli. Optimum aerdynamic design using the Navier- Stkes equatins. In AIAA 97-, 3 5th. Aerspace Sciences Meeting and Exhibit, Ren, Nevada, January 997. American Institute f Aernautics and Astrnautics
11 [] A. Jamesn., L. Martinelli, and N. Pierce Optimum aerdynamic design using the Navier- Stkes equatins. In Theretical Cmputatinal Fluid Dynamics, :3-37, 998. [] G. R. Shubin and P. D. Frank A Cmparisn f the Implicit Gradient Apprach and the Variatinal Apprach t Aerdynamic Design Optimizatin. Being Cmputer Services Reprt AMS-TR-63, April 99. [] J. Reuther and A. Jamesn. Aerdynamic shape ptimizatin f wing and wing-bdy cnfiguratins using cntrl thery. AIAA 95-3, 33rd Aerspace Sciences Meeting and Exibit, Ren, Nevada, January 995. [3] J. Reuther, A. Jamesn, J. Farmer, L. Martinelli, and D. Saunders. Aerdynamic shape ptimizatin f cmplex aircraft cnfiguratins via an adjint frmulatin. AIAA 96-94, AIAA 34th Aerspace Sciences Meeting and Exhibit, Ren, NV, January 996. [] S. Ta asan, G. Kuruvila, and M. D. Salas. Aerdynamic design and ptimizatin in ne sht. AIAA 9-5, 3th Aerspace Sciences Meeting and Exibit, Ren, Nevada, January 99. [] S. Kim,J. J. Alns, and A. Jamesn A Gradient Accuracy Study fr the Adjint-Based Navier-Stkes Design Methd. AIAA 99-99, AIAA 37th. Aerspace Sciences Meeting and Exhibit, Ren, NV, January 999. [] A. Jamesn Slutin f the Euler Equatins fr Tw Dimensinal Transnic Flw By a Multigrid Methd. Applied Mathematics and Cmputatin, 3:37-355, 983. [3] A. Jamesn Studies f Alternative Numerical Optimizatin Methds Applied t the Brachistchrne Prblem. [4] J. Reuther, A. Jamesn, J. J. Alns, M. J Rimlinger, and D. Saunders. Cnstrained multipint aerdynamic shape ptimizatin using an adjint frmulatin and parallel cmputers. AIAA 97-3, AIAA 35th Aerspace Sciences Meeting and Exhibit, Ren, NV, January 997. [5] F. Beux and A. Dervieux. Exact-Gradient Shape Optimizatin f a -D Euler Flw. Finite Elements in Analysis and Design, Vl., 99, 8-3. [6] G. W. Burgreen and O. Baysal. Three- Dimensinal Aerdynamic Shape Optimizatin f Wings Using Discrete Sensitivity Analysis. AIAA Jurnal, Vl. 34, N.9, September 996, pp [7] J. Ellit and J. Peraire. Aerdynamic Design Using Unstructured Meshes. AIAA 96-94, 996. [8] W. K. Andersn and V. Venkatakrishnan Aerdynamic Design Optimizatin n Unstructured Grids with a Cntinuus Adjint Frmulatin. AIAA 96-94, 996. [9] A. Ill, M. Salas, and S. Ta asan. Shape Optimizatin Grvened by the Euler Equatins Using and Adjint Methd. ICASE reprt 93-78, Nvember 993. American Institute f Aernautics and Astrnautics
12 Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E a: Initial Slutin f Krn Airfil b: After 5 Design Iteratins Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E c: After 5 Design Iteratins d: Final Design Figure : Inverse Design f Krn t NACA 64A4 at Fixed C l Grid - 9 x 3, M =.74, C l =.63, α = degrees American Institute f Aernautics and Astrnautics
13 Magnitude f Gradient.. Magnitude f Gradient Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient.6 Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient.8 cnt fdg = 3.6e 3 disc fdg =.397e Design Pint.8 cnt fdg =.73e 3 disc fdg =.74e Design Pint Figure : Adjint Versus Finite Difference Gradients fr Inverse Design f Krn t NACA 64A4 at Fixed C l. Carse Grid - 96 x 6, M =.74, C l =.63 Figure 3: Adjint Versus Finite Difference Gradients fr Inverse Design f Krn t NACA 64A4 at Fixed C l. Medium Grid - 9 x 3, M =.74, C l = Magnitude f Gradient Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg =.44e 3 disc fdg =.49e Design Pint Magnitude f Gradient Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg = 3.88e 3 disc fdg = 3.3e Design Pint Figure 4: Adjint Versus Finite Difference Gradients fr Inverse Design f Krn t NACA 64A4 at Fixed C l. Fine Grid - 56 x 64, M =.74, C l =.63 Figure 5: Adjint Versus Finite Difference Gradients fr Inverse Design f Krn t NACA 64A4 at Fixed C l. Dissipative Cefficients Nt Frzen Medium Grid - 9 x 3, M =.74, C l =.63 3 American Institute f Aernautics and Astrnautics
14 .5.5.e Order.e Order.e 3 Order.e 4 Order.e Order.e Order.e 3 Order.e 4 Order.5.5 Gradient Gradient Design Pints Design Pints Figure 6: Cntinuus Adjint Gradients fr Varying Flw Slver Cnvergence fr the Inverse Design Case Figure 7: Discrete Adjint Gradients fr Varying Flw Slver Cnvergence fr the Inverse Design Case.8.e Order.e Order.e 3 Order.e 4 Order.8.e Order.e Order.e 3 Order.e 4 Order Gradient Gradient Design Pints Design Pints Figure 8: Cntinuus Adjint Gradients fr Varying Adjint Slver Cnvergence fr the Inverse Design Case Figure 9: Discrete Adjint Gradients fr Varying Adjint Slver Cnvergence fr the Inverse Design Case 4 American Institute f Aernautics and Astrnautics
15 Secnd Cstate Variable Third Cstate Variable Cnt Adjint Disc Adjint 4 4 Cntinuus Discrete 8 6 Cntinuus Discrete 8 6 Lg(Errr Value f Cstate Value f Cstate Figure : Cmparisn f Cstate Values Between the Cntinuus and Discrete Adjint Methd fr the Inverse Design f Krn t NACA 64A4 at Fixed C l. Medium Grid - 9 x 3, M =.74, C l = Number f Iteratins Figure : Cnvergence Histry fr the Cntinuus and Discrete Adjint fr the Inverse Design f Krn t NACA 64A4 at Fixed C l. M =.74, C l =.63 5 American Institute f Aernautics and Astrnautics
16 Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E a: Initial Slutin f Krn Airfil b: Final Design Figure : Inverse Design f Krn t NACA 64A4 at Fixed C l Grid - 9 x 3, M =.78, C l =.63, α = degrees Magnitude f Gradient.4.. Magnitude f Gradient.. Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg =.369e 3 disc fdg =.8e 3.4 Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg =.3e 3 disc fdg =.895e Design Pint Design Pint Figure 3: Adjint Versus Finite Difference Gradients fr Inverse Design f Krn t NACA 64A4 at Fixed C l. Carse Grid - 96 x 6, M =.78, C l =.63 Figure 4: Adjint Versus Finite Difference Gradients fr Inverse Design f Krn t NACA 64A4 at Fixed C l. Medium Grid - 9 x 3, M =.78, C l =.63 6 American Institute f Aernautics and Astrnautics
17 Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E 5a: Initial Slutin f NACA 64A4. M =.75, C l =.63, C d =.3, α =.34 degrees 5b: After Design Iteratins. M =.75, C l =.66, C d =.44, α =.8 degrees Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E Cp.E.8E.4E -.E-5 -.4E -.8E -.E -.E 5c: After 4 Design Iteratins M =.75, C l =.66, C d =., α =.8 degrees 5d: After Design Iteratins M =.75, C l =.69, C d =.5, α =.79 degrees Figure 5: Drag Minimizatin f NACA 64A4 at Fixed C l Grid - 9 x 3, M =.75, C l =.63 7 American Institute f Aernautics and Astrnautics
18 Magnitude f Gradient Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg =.9e disc fdg =.75e Magnitude f Gradient 4 6 Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg =.49e disc fdg = 7.577e Design Pint Design Pint Figure 6: Adjint Versus Finite Difference Gradients fr Drag Minimizatin f NACA 64A4 at Fixed C l. Carse Grid - 96 x 6, M =.75, C l =.63 Figure 7: Adjint Versus Finite Difference Gradients fr Drag Minimizatin f NACA 64A4 at Fixed C l. Medium Grid - 9 x 3, M =.75, C l =.63 Magnitude f Gradient 4 6 Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg = 6.4e 3 disc fdg = 5.54e 3 Magnitude f Gradient Cnt Adjint Gradient Disc Adjint Gradient Finite Difference Gradient cnt fdg =.366e disc fdg =.93e Design Pint Figure 8: Adjint Versus Finite Difference Gradients fr Drag Minimizatin f NACA 64A4 at Fixed C l. Fine Grid - 56 x 64, M =.75, C l = Design Pint Figure 9: Adjint Versus Finite Difference Gradients fr Drag Minimizatin f NACA 64A4 at Fixed C l. Dissipative Cefficients Nt Frzen Medium Grid - 9 x 3, M =.75, C l =.63 8 American Institute f Aernautics and Astrnautics
19 .e Order.e Order.e 4 Order.e 6 Order.e Order.e Order.e 4 Order.e 6 Order Gradient 3 4 Gradient Design Pints Design Pints Figure : Cntinuus Adjint Gradients fr Varying Flw Slver Cnvergence fr the Drag Minimizatin Case Figure : Discrete Adjint Gradients fr Varying Flw Slver Cnvergence fr the Drag Minimizatin Case.e Order.e Order.e 3 Order.e 4 Order.e Order.e Order.e 3 Order.e 4 Order Gradient 3 4 Gradient Design Pints Design Pints Figure : Cntinuus Adjint Gradients fr Varying Adjint Slver Cnvergence fr the Drag Minimizatin Case Figure 3: Discrete Adjint Gradients fr Varying Adjint Slver Cnvergence fr the Drag Minimizatin Case 9 American Institute f Aernautics and Astrnautics
20 Cntinuus Adjint Discrete Adjint Objective Functin Design Cycles Figure 4: Cmparisn f Cnvergence f the Objective Functin Between the Cntinuus and Discrete Adjint Methd fr Drag Minimizatin Secnd Cstate Variable Third Cstate Variable 8 6 Cntinuus Discrete 8 6 Cntinuus Discrete Cnt Adjint Disc Adjint Lg(Errr Value f Cstate Value f Cstate Figure 5: Cmparisn f Cstate Values Between the Cntinuus and Discrete Adjint Methd fr Drag Minimizatin f NACA 64A4 at Fixed C l. Medium Grid - 9 x 3, M =.75, C l = Number f Iteratins Figure 6: Cnvergence Histry fr the Cntinuus and Discrete Adjint fr Drag Minimizatin f NACA 64A4 at Fixed C l. M =.75, C l =.63 American Institute f Aernautics and Astrnautics
DESIGN OPTIMIZATION OF HIGH-LIFT CONFIGURATIONS USING A VISCOUS ADJOINT-BASED METHOD
DESIGN OPTIMIZATION OF HIGH-LIFT CONFIGURATIONS USING A VISCOUS ADJOINT-BASED METHOD Sangh Kim Stanfrd University Juan J. Alns Stanfrd University Antny Jamesn Stanfrd University 40th AIAA Aerspace Sciences
More informationReduction of the Adjoint Gradient Formula in the Continuous Limit
Reductin f the Adjint Gradient Frmula in the Cntinuus Limit Antny Jamesn and Sangh Kim Stanfrd University, Stanfrd, CA 9435, U.S.A Abstract We present a new cntinuus adjint methd fr aerdynamic shape ptimizatin
More informationSTUDIES OF THE CONTINUOUS AND DISCRETE ADJOINT APPROACHES TO VISCOUS AUTOMATIC AERODYNAMIC SHAPE OPTIMIZATION
STUIES OF THE CONTINUOUS AN ISCRETE AJOINT APPROACHES TO VISCOUS AUTOMATIC AEROYNAMIC SHAPE OPTIMIZATION Siva K. Nadarajah and Antny Jamesn epartment f Aernautics and Astrnautics Stanfrd University Stanfrd,
More informationDesign Optimization of Multi Element High Lift Configurations. Using a Viscous Continuous Adjoint Method
Design Optimizatin f Multi Element High Lift Cnfiguratins Using a Viscus Cntinuus Adjint Methd Sangh Kim, Juan J. Alns, and Antny Jamesn Stanfrd University, Stanfrd, CA 94305 ABSTRACT An adjint-based Navier-Stkes
More informationAerodynamic Shape Optimization Using the Adjoint Method
Aerdynamic Shape Optimizatin Using the Adjint Methd Antny Jamesn epartment f Aernautics & Astrnautics Stanfrd University Stanfrd, Califrnia 94305 USA Lectures at the Vn Karman Institute, russels Febuary
More informationDesign Optimization of High Lift Configurations Using a Viscous Continuous Adjoint Method
AIAA 2002 0844 Design Optimizatin f High Lift Cnfiguratins Using a Viscus Cntinuus Adjint Methd Sangh Kim, Juan J. Alns, and Antny Jamesn Stanfrd University, Stanfrd, CA 94305 40th AIAA Aerspace Sciences
More informationOptimum Aerodynamic Design using the Navier Stokes Equations
Optimum Aerdynamic esign using the Navier Stkes Equatins A. Jamesn, N.A. Pierce and L. Martinelli, epartment f Mechanical and Aerspace Engineering Princetn University Princetn, New Jersey 08544 USA and
More informationPressure And Entropy Variations Across The Weak Shock Wave Due To Viscosity Effects
Pressure And Entrpy Variatins Acrss The Weak Shck Wave Due T Viscsity Effects OSTAFA A. A. AHOUD Department f athematics Faculty f Science Benha University 13518 Benha EGYPT Abstract:-The nnlinear differential
More information21st AIAA Applied Aerodynamics Conference June 23 26, 2003/Orlando, FL
AIAA 23 3531 Numerical Analysis and Design f Upwind Sails Sriram and Antny Jamesn Dept f Aernautics and Astrnautics Stanfrd University, Stanfrd, CA Margt G Gerritsen Department f Petrleum Engineering Stanfrd
More informationINVERSE PROBLEMS IN AERODYNAMICS AND CONTROL THEORY
INVERSE PROBLEMS IN AERODYNAMICS AND CONTROL THEORY Antny Jamesn Department f Aernautics and Astrnautics Stanfrd University, CA Internatinal Cnference n Cntrl, PDEs and Scientific Cmputing Dedicated t
More informationSupport-Vector Machines
Supprt-Vectr Machines Intrductin Supprt vectr machine is a linear machine with sme very nice prperties. Haykin chapter 6. See Alpaydin chapter 13 fr similar cntent. Nte: Part f this lecture drew material
More informationTHE ROLE OF CFD IN PRELIMINARY AEROSPACE DESIGN
Prceedings f FEDSM 03 4TH ASME JSME Jint Fluids Engineering Cnference Hnlulu, Hawaii USA, July 6-11, 2003 FEDSM2003-45812 THE ROLE OF CFD IN PRELIMINARY AEROSPACE DESIGN Antny Jamesn Department f Aernautics
More informationTHE DISCRETE ADJOINT APPROACH TO AERODYNAMIC SHAPE OPTIMIZATION
THE DISCRETE ADJOINT APPROACH TO AERODYNAMIC SHAPE OPTIMIZATION a dissertatin submitted t the department f aernautics and astrnautics and the cmmittee n graduate studies f stanfrd university in partial
More informationRevision: August 19, E Main Suite D Pullman, WA (509) Voice and Fax
.7.4: Direct frequency dmain circuit analysis Revisin: August 9, 00 5 E Main Suite D Pullman, WA 9963 (509) 334 6306 ice and Fax Overview n chapter.7., we determined the steadystate respnse f electrical
More information39th AIAA Aerospace Sciences Meeting and Exhibit January 8 11, 2001/Reno, NV
AIAA 21 573 Investigatin f Nn-Linear Flutter by a Cupled Aerdynamics and Structural Dynamics Methd Mani Sadeghi and Feng Liu Department f Mechanical and Aerspace Engineering University f Califrnia, Irvine,
More informationModule 4: General Formulation of Electric Circuit Theory
Mdule 4: General Frmulatin f Electric Circuit Thery 4. General Frmulatin f Electric Circuit Thery All electrmagnetic phenmena are described at a fundamental level by Maxwell's equatins and the assciated
More informationModeling the Nonlinear Rheological Behavior of Materials with a Hyper-Exponential Type Function
www.ccsenet.rg/mer Mechanical Engineering Research Vl. 1, N. 1; December 011 Mdeling the Nnlinear Rhelgical Behavir f Materials with a Hyper-Expnential Type Functin Marc Delphin Mnsia Département de Physique,
More informationNumerical Simulation of the Flow Field in a Friction-Type Turbine (Tesla Turbine)
Numerical Simulatin f the Flw Field in a Frictin-Type Turbine (Tesla Turbine) Institute f Thermal Pwerplants Vienna niversity f Technlgy Getreidemarkt 9/313, A-6 Wien Andrés Felipe Rey Ladin Schl f Engineering,
More informationPattern Recognition 2014 Support Vector Machines
Pattern Recgnitin 2014 Supprt Vectr Machines Ad Feelders Universiteit Utrecht Ad Feelders ( Universiteit Utrecht ) Pattern Recgnitin 1 / 55 Overview 1 Separable Case 2 Kernel Functins 3 Allwing Errrs (Sft
More information7 TH GRADE MATH STANDARDS
ALGEBRA STANDARDS Gal 1: Students will use the language f algebra t explre, describe, represent, and analyze number expressins and relatins 7 TH GRADE MATH STANDARDS 7.M.1.1: (Cmprehensin) Select, use,
More informationEDA Engineering Design & Analysis Ltd
EDA Engineering Design & Analysis Ltd THE FINITE ELEMENT METHOD A shrt tutrial giving an verview f the histry, thery and applicatin f the finite element methd. Intrductin Value f FEM Applicatins Elements
More informationOF SIMPLY SUPPORTED PLYWOOD PLATES UNDER COMBINED EDGEWISE BENDING AND COMPRESSION
U. S. FOREST SERVICE RESEARCH PAPER FPL 50 DECEMBER U. S. DEPARTMENT OF AGRICULTURE FOREST SERVICE FOREST PRODUCTS LABORATORY OF SIMPLY SUPPORTED PLYWOOD PLATES UNDER COMBINED EDGEWISE BENDING AND COMPRESSION
More information(1.1) V which contains charges. If a charge density ρ, is defined as the limit of the ratio of the charge contained. 0, and if a force density f
1.0 Review f Electrmagnetic Field Thery Selected aspects f electrmagnetic thery are reviewed in this sectin, with emphasis n cncepts which are useful in understanding magnet design. Detailed, rigrus treatments
More informationIntroductory Thoughts
Flw Similarity By using the Buckingham pi therem, we have reduced the number f independent variables frm five t tw If we wish t run a series f wind-tunnel tests fr a given bdy at a given angle f attack,
More informationFebruary 28, 2013 COMMENTS ON DIFFUSION, DIFFUSIVITY AND DERIVATION OF HYPERBOLIC EQUATIONS DESCRIBING THE DIFFUSION PHENOMENA
February 28, 2013 COMMENTS ON DIFFUSION, DIFFUSIVITY AND DERIVATION OF HYPERBOLIC EQUATIONS DESCRIBING THE DIFFUSION PHENOMENA Mental Experiment regarding 1D randm walk Cnsider a cntainer f gas in thermal
More informationHomology groups of disks with holes
Hmlgy grups f disks with hles THEOREM. Let p 1,, p k } be a sequence f distinct pints in the interir unit disk D n where n 2, and suppse that fr all j the sets E j Int D n are clsed, pairwise disjint subdisks.
More informationA Few Basic Facts About Isothermal Mass Transfer in a Binary Mixture
Few asic Facts but Isthermal Mass Transfer in a inary Miture David Keffer Department f Chemical Engineering University f Tennessee first begun: pril 22, 2004 last updated: January 13, 2006 dkeffer@utk.edu
More informationLead/Lag Compensator Frequency Domain Properties and Design Methods
Lectures 6 and 7 Lead/Lag Cmpensatr Frequency Dmain Prperties and Design Methds Definitin Cnsider the cmpensatr (ie cntrller Fr, it is called a lag cmpensatr s K Fr s, it is called a lead cmpensatr Ntatin
More information22.54 Neutron Interactions and Applications (Spring 2004) Chapter 11 (3/11/04) Neutron Diffusion
.54 Neutrn Interactins and Applicatins (Spring 004) Chapter (3//04) Neutrn Diffusin References -- J. R. Lamarsh, Intrductin t Nuclear Reactr Thery (Addisn-Wesley, Reading, 966) T study neutrn diffusin
More informationComputational modeling techniques
Cmputatinal mdeling techniques Lecture 4: Mdel checing fr ODE mdels In Petre Department f IT, Åb Aademi http://www.users.ab.fi/ipetre/cmpmd/ Cntent Stichimetric matrix Calculating the mass cnservatin relatins
More information7.0 Heat Transfer in an External Laminar Boundary Layer
7.0 Heat ransfer in an Eternal Laminar Bundary Layer 7. Intrductin In this chapter, we will assume: ) hat the fluid prperties are cnstant and unaffected by temperature variatins. ) he thermal & mmentum
More informationBuilding to Transformations on Coordinate Axis Grade 5: Geometry Graph points on the coordinate plane to solve real-world and mathematical problems.
Building t Transfrmatins n Crdinate Axis Grade 5: Gemetry Graph pints n the crdinate plane t slve real-wrld and mathematical prblems. 5.G.1. Use a pair f perpendicular number lines, called axes, t define
More informationMath 302 Learning Objectives
Multivariable Calculus (Part I) 13.1 Vectrs in Three-Dimensinal Space Math 302 Learning Objectives Plt pints in three-dimensinal space. Find the distance between tw pints in three-dimensinal space. Write
More informationBootstrap Method > # Purpose: understand how bootstrap method works > obs=c(11.96, 5.03, 67.40, 16.07, 31.50, 7.73, 11.10, 22.38) > n=length(obs) >
Btstrap Methd > # Purpse: understand hw btstrap methd wrks > bs=c(11.96, 5.03, 67.40, 16.07, 31.50, 7.73, 11.10, 22.38) > n=length(bs) > mean(bs) [1] 21.64625 > # estimate f lambda > lambda = 1/mean(bs);
More information, which yields. where z1. and z2
The Gaussian r Nrmal PDF, Page 1 The Gaussian r Nrmal Prbability Density Functin Authr: Jhn M Cimbala, Penn State University Latest revisin: 11 September 13 The Gaussian r Nrmal Prbability Density Functin
More informationA Simple Set of Test Matrices for Eigenvalue Programs*
Simple Set f Test Matrices fr Eigenvalue Prgrams* By C. W. Gear** bstract. Sets f simple matrices f rder N are given, tgether with all f their eigenvalues and right eigenvectrs, and simple rules fr generating
More informationOn Boussinesq's problem
Internatinal Jurnal f Engineering Science 39 (2001) 317±322 www.elsevier.cm/lcate/ijengsci On Bussinesq's prblem A.P.S. Selvadurai * Department f Civil Engineering and Applied Mechanics, McGill University,
More informationAircraft Performance - Drag
Aircraft Perfrmance - Drag Classificatin f Drag Ntes: Drag Frce and Drag Cefficient Drag is the enemy f flight and its cst. One f the primary functins f aerdynamicists and aircraft designers is t reduce
More informationChapter 3: Cluster Analysis
Chapter 3: Cluster Analysis } 3.1 Basic Cncepts f Clustering 3.1.1 Cluster Analysis 3.1. Clustering Categries } 3. Partitining Methds 3..1 The principle 3.. K-Means Methd 3..3 K-Medids Methd 3..4 CLARA
More informationLeast Squares Optimal Filtering with Multirate Observations
Prc. 36th Asilmar Cnf. n Signals, Systems, and Cmputers, Pacific Grve, CA, Nvember 2002 Least Squares Optimal Filtering with Multirate Observatins Charles W. herrien and Anthny H. Hawes Department f Electrical
More informationAP Statistics Notes Unit Two: The Normal Distributions
AP Statistics Ntes Unit Tw: The Nrmal Distributins Syllabus Objectives: 1.5 The student will summarize distributins f data measuring the psitin using quartiles, percentiles, and standardized scres (z-scres).
More informationNumerical Simulation of the Thermal Resposne Test Within the Comsol Multiphysics Environment
Presented at the COMSOL Cnference 2008 Hannver University f Parma Department f Industrial Engineering Numerical Simulatin f the Thermal Respsne Test Within the Cmsl Multiphysics Envirnment Authr : C. Crradi,
More informationMATHEMATICS SYLLABUS SECONDARY 5th YEAR
Eurpean Schls Office f the Secretary-General Pedaggical Develpment Unit Ref. : 011-01-D-8-en- Orig. : EN MATHEMATICS SYLLABUS SECONDARY 5th YEAR 6 perid/week curse APPROVED BY THE JOINT TEACHING COMMITTEE
More informationPure adaptive search for finite global optimization*
Mathematical Prgramming 69 (1995) 443-448 Pure adaptive search fr finite glbal ptimizatin* Z.B. Zabinskya.*, G.R. Wd b, M.A. Steel c, W.P. Baritmpa c a Industrial Engineering Prgram, FU-20. University
More informationStudy Group Report: Plate-fin Heat Exchangers: AEA Technology
Study Grup Reprt: Plate-fin Heat Exchangers: AEA Technlgy The prblem under study cncerned the apparent discrepancy between a series f experiments using a plate fin heat exchanger and the classical thery
More informationHow do scientists measure trees? What is DBH?
Hw d scientists measure trees? What is DBH? Purpse Students develp an understanding f tree size and hw scientists measure trees. Students bserve and measure tree ckies and explre the relatinship between
More informationCOMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING 85 (1991) NORTH-HOLLAND
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING 85 (1991) 207-217 NORTH-HOLLAND ON THE DOWNSTREAM BOUNDARY CONDITIONS FOR THE VORTICITY-STREAM FUNCTION FORMULATION OF TWO-DIMENSIONAL INCOMPRESSIBLE
More informationChapter 3 Kinematics in Two Dimensions; Vectors
Chapter 3 Kinematics in Tw Dimensins; Vectrs Vectrs and Scalars Additin f Vectrs Graphical Methds (One and Tw- Dimensin) Multiplicatin f a Vectr b a Scalar Subtractin f Vectrs Graphical Methds Adding Vectrs
More informationON THE EFFECTIVENESS OF POROSITY ON UNSTEADY COUETTE FLOW AND HEAT TRANSFER BETWEEN PARALLEL POROUS PLATES WITH EXPONENTIAL DECAYING PRESSURE GRADIENT
17 Kragujevac J. Sci. 8 (006) 17-4. ON THE EFFECTIVENESS OF POROSITY ON UNSTEADY COUETTE FLOW AND HEAT TRANSFER BETWEEN PARALLEL POROUS PLATES WITH EXPONENTIAL DECAYING PRESSURE GRADIENT Hazem Ali Attia
More informationCOMP 551 Applied Machine Learning Lecture 11: Support Vector Machines
COMP 551 Applied Machine Learning Lecture 11: Supprt Vectr Machines Instructr: (jpineau@cs.mcgill.ca) Class web page: www.cs.mcgill.ca/~jpineau/cmp551 Unless therwise nted, all material psted fr this curse
More information3.4 Shrinkage Methods Prostate Cancer Data Example (Continued) Ridge Regression
3.3.4 Prstate Cancer Data Example (Cntinued) 3.4 Shrinkage Methds 61 Table 3.3 shws the cefficients frm a number f different selectin and shrinkage methds. They are best-subset selectin using an all-subsets
More informationCOMP 551 Applied Machine Learning Lecture 9: Support Vector Machines (cont d)
COMP 551 Applied Machine Learning Lecture 9: Supprt Vectr Machines (cnt d) Instructr: Herke van Hf (herke.vanhf@mail.mcgill.ca) Slides mstly by: Class web page: www.cs.mcgill.ca/~hvanh2/cmp551 Unless therwise
More informationSlide04 (supplemental) Haykin Chapter 4 (both 2nd and 3rd ed): Multi-Layer Perceptrons
Slide04 supplemental) Haykin Chapter 4 bth 2nd and 3rd ed): Multi-Layer Perceptrns CPSC 636-600 Instructr: Ynsuck Che Heuristic fr Making Backprp Perfrm Better 1. Sequential vs. batch update: fr large
More informationFree Vibrations of Catenary Risers with Internal Fluid
Prceeding Series f the Brazilian Sciety f Applied and Cmputatinal Mathematics, Vl. 4, N. 1, 216. Trabalh apresentad n DINCON, Natal - RN, 215. Prceeding Series f the Brazilian Sciety f Cmputatinal and
More informationMath Foundations 20 Work Plan
Math Fundatins 20 Wrk Plan Units / Tpics 20.8 Demnstrate understanding f systems f linear inequalities in tw variables. Time Frame December 1-3 weeks 6-10 Majr Learning Indicatrs Identify situatins relevant
More informationAnalysis of Hydrodynamics and Heat Transfer in a Thin Liquid Film Flowing Over a Rotating Disk by the Integral Method
S. Basu B. M. Cetegen 1 Fellw ASME Mechanical Engineering Department, University f Cnnecticut, Strrs, CT 06269-3139 Analysis f Hydrdynamics and Heat Transfer in a Thin Liquid Film Flwing Over a Rtating
More informationFloating Point Method for Solving Transportation. Problems with Additional Constraints
Internatinal Mathematical Frum, Vl. 6, 20, n. 40, 983-992 Flating Pint Methd fr Slving Transprtatin Prblems with Additinal Cnstraints P. Pandian and D. Anuradha Department f Mathematics, Schl f Advanced
More informationRay tracing equations in transversely isotropic media Cosmin Macesanu and Faruq Akbar, Seimax Technologies, Inc.
Ray tracing equatins in transversely istrpic media Csmin Macesanu and Faruq Akbar, Seimax Technlgies, Inc. SUMMARY We discuss a simple, cmpact apprach t deriving ray tracing equatins in transversely istrpic
More informationthe results to larger systems due to prop'erties of the projection algorithm. First, the number of hidden nodes must
M.E. Aggune, M.J. Dambrg, M.A. El-Sharkawi, R.J. Marks II and L.E. Atlas, "Dynamic and static security assessment f pwer systems using artificial neural netwrks", Prceedings f the NSF Wrkshp n Applicatins
More informationFall 2013 Physics 172 Recitation 3 Momentum and Springs
Fall 03 Physics 7 Recitatin 3 Mmentum and Springs Purpse: The purpse f this recitatin is t give yu experience wrking with mmentum and the mmentum update frmula. Readings: Chapter.3-.5 Learning Objectives:.3.
More information2D Compressible Viscous-Flow Solver on. Unstructured Meshes with Linear and. Quadratic Reconstruction of Convective. Fluxes
2D Cmpressible Viscus-Flw Slver n Unstructured Meshes with Linear and Quadratic Recnstructin f Cnvective Fluxes Interim Prject Reprt Master f Technlgy By Mhamed Yusuf A U ( Rll N 0440304 ) DEPARTMENT OF
More informationLHS Mathematics Department Honors Pre-Calculus Final Exam 2002 Answers
LHS Mathematics Department Hnrs Pre-alculus Final Eam nswers Part Shrt Prblems The table at the right gives the ppulatin f Massachusetts ver the past several decades Using an epnential mdel, predict the
More informationTHE TOPOLOGY OF SURFACE SKIN FRICTION AND VORTICITY FIELDS IN WALL-BOUNDED FLOWS
THE TOPOLOGY OF SURFACE SKIN FRICTION AND VORTICITY FIELDS IN WALL-BOUNDED FLOWS M.S. Chng Department f Mechanical Engineering The University f Melburne Victria 3010 AUSTRALIA min@unimelb.edu.au J.P. Mnty
More informationChapter 2 GAUSS LAW Recommended Problems:
Chapter GAUSS LAW Recmmended Prblems: 1,4,5,6,7,9,11,13,15,18,19,1,7,9,31,35,37,39,41,43,45,47,49,51,55,57,61,6,69. LCTRIC FLUX lectric flux is a measure f the number f electric filed lines penetrating
More informationCHAPTER 3 INEQUALITIES. Copyright -The Institute of Chartered Accountants of India
CHAPTER 3 INEQUALITIES Cpyright -The Institute f Chartered Accuntants f India INEQUALITIES LEARNING OBJECTIVES One f the widely used decisin making prblems, nwadays, is t decide n the ptimal mix f scarce
More informationIntroduction: A Generalized approach for computing the trajectories associated with the Newtonian N Body Problem
A Generalized apprach fr cmputing the trajectries assciated with the Newtnian N Bdy Prblem AbuBar Mehmd, Syed Umer Abbas Shah and Ghulam Shabbir Faculty f Engineering Sciences, GIK Institute f Engineering
More informationThe Sputtering Problem James A Glackin, James V. Matheson
The Sputtering Prblem James A Glackin, James V. Mathesn I prpse t cnsider first the varius elements f the subject, next its varius parts r sectins, and finally the whle in its internal structure. In ther
More informationA Matrix Representation of Panel Data
web Extensin 6 Appendix 6.A A Matrix Representatin f Panel Data Panel data mdels cme in tw brad varieties, distinct intercept DGPs and errr cmpnent DGPs. his appendix presents matrix algebra representatins
More informationDetermining the Accuracy of Modal Parameter Estimation Methods
Determining the Accuracy f Mdal Parameter Estimatin Methds by Michael Lee Ph.D., P.E. & Mar Richardsn Ph.D. Structural Measurement Systems Milpitas, CA Abstract The mst cmmn type f mdal testing system
More informationModelling of NOLM Demultiplexers Employing Optical Soliton Control Pulse
Micwave and Optical Technlgy Letters, Vl. 1, N. 3, 1999. pp. 05-08 Mdelling f NOLM Demultiplexers Emplying Optical Slitn Cntrl Pulse Z. Ghassemly, C. Y. Cheung & A. K. Ray Electrnics Research Grup, Schl
More informationEXPERIMENTAL STUDY ON DISCHARGE COEFFICIENT OF OUTFLOW OPENING FOR PREDICTING CROSS-VENTILATION FLOW RATE
EXPERIMENTAL STUD ON DISCHARGE COEFFICIENT OF OUTFLOW OPENING FOR PREDICTING CROSS-VENTILATION FLOW RATE Tmnbu Gt, Masaaki Ohba, Takashi Kurabuchi 2, Tmyuki End 3, shihik Akamine 4, and Tshihir Nnaka 2
More informationWRITING THE REPORT. Organizing the report. Title Page. Table of Contents
WRITING THE REPORT Organizing the reprt Mst reprts shuld be rganized in the fllwing manner. Smetime there is a valid reasn t include extra chapters in within the bdy f the reprt. 1. Title page 2. Executive
More informationThe Destabilization of Rossby Normal Modes by Meridional Baroclinic Shear
The Destabilizatin f Rssby Nrmal Mdes by Meridinal Barclinic Shear by Jseph Pedlsky Wds Hle Oceangraphic Institutin Wds Hle, MA 0543 Abstract The Rssby nrmal mdes f a tw-layer fluid in a meridinal channel
More information3D FE Modeling Simulation of Cold Rotary Forging with Double Symmetry Rolls X. H. Han 1, a, L. Hua 1, b, Y. M. Zhao 1, c
Materials Science Frum Online: 2009-08-31 ISSN: 1662-9752, Vls. 628-629, pp 623-628 di:10.4028/www.scientific.net/msf.628-629.623 2009 Trans Tech Publicatins, Switzerland 3D FE Mdeling Simulatin f Cld
More informationLyapunov Stability Stability of Equilibrium Points
Lyapunv Stability Stability f Equilibrium Pints 1. Stability f Equilibrium Pints - Definitins In this sectin we cnsider n-th rder nnlinear time varying cntinuus time (C) systems f the frm x = f ( t, x),
More informationEFFICIENT SOLUTIONS OF THE EULER AND NAVIER-STOKES EQUATIONS FOR EXTERNAL FLOWS
Scientific Bulletin f the Plitehnica University f Timisara Transactins n Mechanics Special issue The 6 th Internatinal Cnference n Hydraulic Machinery and Hydrdynamics Timisara, Rmania, Octber -, 4 EFFICIENT
More informationDifferentiation Applications 1: Related Rates
Differentiatin Applicatins 1: Related Rates 151 Differentiatin Applicatins 1: Related Rates Mdel 1: Sliding Ladder 10 ladder y 10 ladder 10 ladder A 10 ft ladder is leaning against a wall when the bttm
More informationFIELD QUALITY IN ACCELERATOR MAGNETS
FIELD QUALITY IN ACCELERATOR MAGNETS S. Russenschuck CERN, 1211 Geneva 23, Switzerland Abstract The field quality in the supercnducting magnets is expressed in terms f the cefficients f the Furier series
More informationNUMERICAL SIMULATION OF CHLORIDE DIFFUSION IN REINFORCED CONCRETE STRUCTURES WITH CRACKS
VIII Internatinal Cnference n Fracture Mechanics f Cnete and Cnete Structures FraMCS-8 J.G.M. Van Mier, G. Ruiz, C. Andrade, R.C. Yu and X.X. Zhang (Eds) NUMERICAL SIMULATION OF CHLORIDE DIFFUSION IN REINFORCED
More informationCOASTAL ENGINEERING Chapter 2
CASTAL ENGINEERING Chapter 2 GENERALIZED WAVE DIFFRACTIN DIAGRAMS J. W. Jhnsn Assciate Prfessr f Mechanical Engineering University f Califrnia Berkeley, Califrnia INTRDUCTIN Wave diffractin is the phenmenn
More informationAdmissibility Conditions and Asymptotic Behavior of Strongly Regular Graphs
Admissibility Cnditins and Asympttic Behavir f Strngly Regular Graphs VASCO MOÇO MANO Department f Mathematics University f Prt Oprt PORTUGAL vascmcman@gmailcm LUÍS ANTÓNIO DE ALMEIDA VIEIRA Department
More informationEHed of Curvature on the Temperature Profiles
PROC. OF THE OKLA. ACAD. OF SCI. FOR 1967 EHed f Curvature n the Temperature Prfiles in Cnduding Spines J. E. FRANCIS add R. V. KASER, University f Oklahma, Nrman and GORDON SCOFIELD, University f Missuri,
More informationThe blessing of dimensionality for kernel methods
fr kernel methds Building classifiers in high dimensinal space Pierre Dupnt Pierre.Dupnt@ucluvain.be Classifiers define decisin surfaces in sme feature space where the data is either initially represented
More informationinitially lcated away frm the data set never win the cmpetitin, resulting in a nnptimal nal cdebk, [2] [3] [4] and [5]. Khnen's Self Organizing Featur
Cdewrd Distributin fr Frequency Sensitive Cmpetitive Learning with One Dimensinal Input Data Aristides S. Galanpuls and Stanley C. Ahalt Department f Electrical Engineering The Ohi State University Abstract
More informationAerodynamic Separability in Tip Speed Ratio and Separability in Wind Speed- a Comparison
Jurnal f Physics: Cnference Series OPEN ACCESS Aerdynamic Separability in Tip Speed Rati and Separability in Wind Speed- a Cmparisn T cite this article: M L Gala Sants et al 14 J. Phys.: Cnf. Ser. 555
More informationA New Evaluation Measure. J. Joiner and L. Werner. The problems of evaluation and the needed criteria of evaluation
III-l III. A New Evaluatin Measure J. Jiner and L. Werner Abstract The prblems f evaluatin and the needed criteria f evaluatin measures in the SMART system f infrmatin retrieval are reviewed and discussed.
More informationReview Problems 3. Four FIR Filter Types
Review Prblems 3 Fur FIR Filter Types Fur types f FIR linear phase digital filters have cefficients h(n fr 0 n M. They are defined as fllws: Type I: h(n = h(m-n and M even. Type II: h(n = h(m-n and M dd.
More informationLecture 17: Free Energy of Multi-phase Solutions at Equilibrium
Lecture 17: 11.07.05 Free Energy f Multi-phase Slutins at Equilibrium Tday: LAST TIME...2 FREE ENERGY DIAGRAMS OF MULTI-PHASE SOLUTIONS 1...3 The cmmn tangent cnstructin and the lever rule...3 Practical
More informationChE 471: LECTURE 4 Fall 2003
ChE 47: LECTURE 4 Fall 003 IDEL RECTORS One f the key gals f chemical reactin engineering is t quantify the relatinship between prductin rate, reactr size, reactin kinetics and selected perating cnditins.
More informationRotating Radial Duct
Internatinl Jurnal f Rtating Machinery 1995, Vl. 2, N. 1, pp. 51-58 Reprints available directly frm the publisher Phtcpying permitted by license nly (C) 1995 OPA (Overseas Publishers Assciatin) Amsterdam
More informationAPPLICATION OF THE BRATSETH SCHEME FOR HIGH LATITUDE INTERMITTENT DATA ASSIMILATION USING THE PSU/NCAR MM5 MESOSCALE MODEL
JP2.11 APPLICATION OF THE BRATSETH SCHEME FOR HIGH LATITUDE INTERMITTENT DATA ASSIMILATION USING THE PSU/NCAR MM5 MESOSCALE MODEL Xingang Fan * and Jeffrey S. Tilley University f Alaska Fairbanks, Fairbanks,
More informationFigure 1a. A planar mechanism.
ME 5 - Machine Design I Fall Semester 0 Name f Student Lab Sectin Number EXAM. OPEN BOOK AND CLOSED NOTES. Mnday, September rd, 0 Write n ne side nly f the paper prvided fr yur slutins. Where necessary,
More informationEffects of piezo-viscous dependency on squeeze film between circular plates: Couple Stress fluid model
Turkish Jurnal f Science & Technlgy Vlume 9(1), 97-103, 014 Effects f piez-viscus dependency n squeeze film between circular plates: Cuple Stress fluid mdel Abstract U. P. SINGH Ansal Technical Campus,
More information1 The limitations of Hartree Fock approximation
Chapter: Pst-Hartree Fck Methds - I The limitatins f Hartree Fck apprximatin The n electrn single determinant Hartree Fck wave functin is the variatinal best amng all pssible n electrn single determinants
More informationANSWER KEY FOR MATH 10 SAMPLE EXAMINATION. Instructions: If asked to label the axes please use real world (contextual) labels
ANSWER KEY FOR MATH 10 SAMPLE EXAMINATION Instructins: If asked t label the axes please use real wrld (cntextual) labels Multiple Chice Answers: 0 questins x 1.5 = 30 Pints ttal Questin Answer Number 1
More informationAdvanced Heat and Mass Transfer by Amir Faghri, Yuwen Zhang, and John R. Howell
6.5 Natural Cnvectin in Enclsures Enclsures are finite spaces bunded by walls and filled with fluid. Natural cnvectin in enclsures, als knwn as internal cnvectin, takes place in rms and buildings, furnaces,
More informationWAVE RESISTANCE AND LIFT ON CYLINDERS BY A COUPLED ELEMENT TECHNIQUE. R. Eatock Taylor and G.X. Wu
E Lab.. Sh,essbu&kunde Technic. r. WAVE RESISTANCE AND LIFT ON CYLINDERS BY A COUPLED ELEMENT TECHNIQUE R. Eatck Taylr and G.X. Wu Lndn Centre fr Marine Technlgy Department f Mechanical Engineering University
More informationMass transport with varying diffusion- and solubility coefficient through a catalytic membrane layer
Mass transprt with varying diffusin- and slubility cefficient thrugh a catalytic membrane layer Prceedings f Eurpean Cngress f Chemical Engineering (ECCE-6) Cpenhagen, 6-0 September 007 Mass transprt with
More informationKinematic transformation of mechanical behavior Neville Hogan
inematic transfrmatin f mechanical behavir Neville Hgan Generalized crdinates are fundamental If we assume that a linkage may accurately be described as a cllectin f linked rigid bdies, their generalized
More informationComparison of two variable parameter Muskingum methods
Extreme Hydrlgical Events: Precipitatin, Flds and Drughts (Prceedings f the Ykhama Sympsium, July 1993). IAHS Publ. n. 213, 1993. 129 Cmparisn f tw variable parameter Muskingum methds M. PERUMAL Department
More information