Utafiti huu unalenga kutumia kwa ufanisi zaidi eneo la ndani la uso katika vibadilisha joto kompakt vyenye nyuso ndogo za vipindi vitatu (TPMS) vinavyoweza kutengenezwa kwa utengenezaji wa nyongeza. Watafiti hawajachukulia baffles kama vizuizi vinavyoongeza turbulence pekee, bali kama vipengele vya usimamizi wa mtiririko vinavyoelekeza mtiririko wa moto na baridi mbali na njia fupi zenye upinzani mdogo, kupunguza maeneo yaliyotuama na kupanga namna ambavyo mitiririko miwili hukutana kwenye ukuta wa pamoja wa TPMS.

Katika mbinu ya uboreshaji ya viwango viwili iliyopendekezwa, mzunguko wa nje hutafuta urefu wa njia ya mtiririko ya uwakilishi unaolingana na upotevu wa shinikizo unaoruhusiwa. Mzunguko wa ndani huongoza algoriti ya kijenetiki kwa kutumia maarifa ya awali ya kimwili yenye gharama ndogo ya kukokotoa kuhusu urefu wa njia, mikunjo iliyo karibu na ingizo, usawaziko wa upana wa njia, maeneo yaliyokufa na ulinganifu wa counterflow wa moto-baridi. Uchambuzi wa gharama kubwa wa COMSOL CFD hautumiki kuchanganua wagombea wote, bali kuthibitisha upotevu wa shinikizo wa idadi ndogo ya miundo iliyopita uchujaji wa awali na kupanga viwango vya uhamishaji joto.

Katika hali ya uwakilishi ya aina ya P yenye kasi ya ingizo ya 1 m/s na kikomo cha upotevu wa shinikizo cha 20 kPa, mbinu iliyopendekezwa ilifikia uhamishaji joto wa 37,96 W kwa tathmini 10 tu za CFD. Chini ya kikomo kile kile, algoriti ya kawaida ya kijenetiki inayofanya kazi moja kwa moja na CFD ilipata 28,97 W kwa takribani tathmini 100, huku kiini cha TPMS bila baffle kikitoa 21,63 W. Hivyo muundo uliopendekezwa ulitoa uhamishaji joto wa juu kwa %75,5 ikilinganishwa na rejea isiyo na baffle, na takribani %31 juu ya matokeo bora zaidi yanayotekelezeka ya algoriti ya moja kwa moja ya kijenetiki.

Katika miundo sita tofauti ya TPMS na hali za uendeshaji, ongezeko la uhamishaji joto la miundo bora iliyotimiza kikomo cha upotevu wa shinikizo lilitofautiana kati ya %24,4 na %78,6 ikilinganishwa na miundo yake sawia isiyo na baffle. Hata hivyo, ulinganisho kamili uliodhibitiwa na algoriti ya moja kwa moja ya kijenetiki ulifanywa tu katika hali ya uwakilishi ya aina ya P ya 20 kPa. Matokeo yanategemea simulation za nambari; hakuna prototype iliyotengenezwa kwa additive manufacturing, hakuna jaribio la experimental heat exchanger, wala uthibitishaji wa surface roughness, fouling, manufacturing tolerance au structural strength.

Kwa nini eneo kubwa la uso pekee halitoshi katika TPMS heat exchangers?

Triply periodic minimal surfaces ni geometria zinazojirudia katika mielekeo mitatu angani, zenye curvature laini na zinazounda mitandao miwili ya mtiririko isiyogusana. Sifa hizi zinafanya TPMS structures kuvutia kwa compact heat exchangers kwa sababu ya high surface-area density, continuous flow channels na complex geometries zinazofaa additive manufacturing.

Hata hivyo, kuwa na total internal surface area kubwa hakumaanishi kwamba uso wote huo unashiriki kwa ufanisi katika heat transfer. Ndani ya open connected channel networks mbili, fluid inaweza kuchagua njia yenye resistance ndogo zaidi kutoka inlet hadi outlet. Katika hali hii, high-speed short-circuit paths zinaweza kutokea katika baadhi ya maeneo, huku maeneo mengine yakibaki na low velocity, underfeeding au stagnation. Uso wa heat exchanger unaopatikana kimwili unaweza kutumiwa vibaya kwa upande wa thermal kwa sababu haupitishwi na flow kwa kiwango cha kutosha.

Ulinganisho wa Figure 1 unaonyesha mechanism hii wazi. Katika structure isiyo na baffle, streamlines hukusanyika katika corridors fulani kati ya inlet na outlet, huku sehemu kubwa za core zikibaki katika low speed. Baffles zinapoongezwa, njia za hot na cold flows hurefuka, flow huelekezwa kwenye sehemu tofauti za core na baadhi ya low-speed regions huanzishwa upya.

Jukumu la baffles linatofautianaje na heat exchangers za kawaida?

Katika shell-and-tube heat exchangers, baffles hutumiwa hasa kuelekeza flow kwa kuvuka, kuongeza turbulence na kuvuruga boundary layer. Katika TPMS channels, kwa kuwa fluid inaweza kubadili route kati ya pores zilizounganishwa, kazi kuu ya baffle ni kusimamia flow path zaidi kuliko kuzalisha turbulence moja kwa moja.

Katika njia ya watafiti, baffle hufanya kazi zifuatazo:

• Kuzuia njia fupi kupita kiasi na zenye resistance ndogo kati ya inlet na outlet,
• Kuelekeza flow kwenye maeneo ya TPMS yasiyotumiwa vya kutosha,
• Kuzuia local contractions na expansions zisizidi kupita kiasi,
• Kupunguza unnecessary pressure loss inayotokana na sharp turns karibu na inlet,
• Kufanya hot na cold flows ziende kwa ulinganifu bora zaidi wa counterflow kando ya common wall,
• Kuamilisha surface kubwa zaidi kwa upande wa thermal ndani ya fixed pumping power au pressure-drop limit.

Tatizo kuu la optimization limefafanuliwaje?

Baffle layouts kwenye cold na hot sides zinaonyeshwa kwa \(B^c\) na \(B^h\) mtawalia. Lengo ni kuongeza heat transfer iliyokokotolewa na CFD hadi kiwango cha juu bila kuzidi upper pressure-drop limit:

\[ \max_{B^c,B^h} J_{\mathrm{CFD}}\left(B^c,B^h\right) \quad \text{koşuluyla} \quad \max\left(\Delta p_c,\Delta p_h\right)\leq\Delta p_{\max} \]

Hapa \(J_{\mathrm{CFD}}\) ni thermal objective function inayotokana na full flow na heat-transfer solution; \(\Delta p_c\) na \(\Delta p_h\) ni pressure drops kwenye cold na hot sides; na \(\Delta p_{\max}\) ni maximum allowable pressure drop kwa flow circuits zote mbili.

Kigezo cha mwisho cha uchaguzi katika utafiti ni wastani wa heat transfer iliyokokotolewa kwa pande mbili:

\[ Q=\frac{Q_c+Q_h}{2} \]

\(Q_c\) na \(Q_h\) ni heat-transfer values zilizokokotolewa kutoka cold na hot flow mtawalia. Katika steady-state solution yenye adiabatic outer surfaces, thamani hizi mbili zinatarajiwa kuwa karibu. Tofauti ndogo kati yake huakisi numerical energy-balance residual.

Kwa nini pressure drop imetafsiriwa kama hydraulic budget?

Katika ulinganisho wote volumetric flow rates ziliwekwa constant. Pumping power ya flow circuit moja inaelezwa kwa uhusiano:

\[ W_p=\frac{\Delta p\,\dot{V}}{\eta_p} \]

Hapa \(W_p\) ni required pumping power; \(\Delta p\) ni pressure drop; \(\dot{V}\) ni volumetric flow rate; na \(\eta_p\) ni pump efficiency. Flow rate na pump efficiency zikiwa constant, kuweka kikomo cha pressure drop ni sawa na kuweka upper bound kwa allowable pumping power.

Total pumping power ya flow circuits mbili ni:

\[ W_{p,\mathrm{toplam}}=\frac{\Delta p_c\dot{V}_c}{\eta_c}+\frac{\Delta p_h\dot{V}_h}{\eta_h} \]

Kwa hiyo utafiti hautafuti highest possible heat transfer bila kuzingatia pressure drop. Swali halisi ni useful thermal capacity kiasi gani kinaweza kupatikana ndani ya fixed hydraulic allowance.

Hata hivyo, actual pumping power ya designs zote si sawa. Pressure-drop constraint huweka tu common upper hydraulic allowance. Baadhi ya designs hutumia sehemu ndogo ya allowance hiyo, nyingine hutumia karibu limit yote.

Geometry ya baffle imewakilishwaje kwa nambari?

Kila flow channel imegawanywa katika grid ya structured Cartesian cells. Kuwepo au kutokuwepo kwa baffle katika horizontal na vertical cell boundaries kumeonyeshwa kwa binary matrices:

\[ B^Q=\left(B^{H,Q},B^{V,Q}\right) \]

\[ B^{H,Q}\in\{0,1\}^{N_x\times(N_y+1)}, \qquad B^{V,Q}\in\{0,1\}^{(N_x+1)\times N_y} \]

\(Q\) inawakilisha cold au hot channel; \(B^{H,Q}\) horizontal baffle segments; na \(B^{V,Q}\) vertical baffle segments. Thamani moja inaonyesha kwamba kuna baffle kwenye boundary husika, huku sifuri ikimaanisha njia iko wazi.

Kutoka kwenye wall layout hii, accessibility field \(C^Q\) ilitengenezwa kuonyesha cells zinazoweza kufikiwa na fluid. Open cells zilifafanuliwa kama graph nodes, na neighboring cells zisizo na baffle kati yake kama graph edges. Hivyo kila baffle layout ilibadilishwa kuwa connectivity problem yenye accessible paths kutoka inlet hadi outlet.

Kwa nini two-level optimization ilitumika?

Kutatua kila genetic-algorithm candidate moja kwa moja kwa full CFD kungeleta computational cost kubwa sana. Utafiti ulijaribu kutatua hili kwa kugawa kazi kati ya outer na inner loops.

Outer loop: kutafuta path length inayolingana na pressure limit

Outer loop hurekebisha target representative path length \(L_{\mathrm{tar}}\) kwa binary search. Kwa kuwa hydraulic resistance kwa kawaida hutarajiwa kuongezeka kadiri flow path inavyorefuka, path length ilitumika kama cheap approximate control variable ya pressure drop.

Kwa kila trial path length, inner optimization huzalisha candidate layout, candidate hutatuliwa kwa COMSOL, na maximum pressure drop ya pande mbili hulinganishwa na limit. Candidate akiwa chini ya limit, longer path hujaribiwa; akizidi limit, path hupunguzwa. Tofauti kati ya lower na upper path bounds ikishuka chini ya grid unit moja, acceptable target length huamuliwa.

Inner loop: kutafuta baffle placement kwa length iliyowekwa

Inner loop hutafuta baffles kwenye cold na hot sides kwa genetic algorithm. Hata hivyo, candidates hawapewi alama kwa CFD bali kwa priors tano zinazoakisi kwa makadirio jinsi flow inavyopaswa kutenda kimwili. Kwa njia hii geometric candidates wengi huchujwa haraka, na CFD hutumiwa tu kwa small promising candidate set.

Katika workflow ya Figure 2, pressure-drop limit na inlet-outlet conditions hufafanuliwa kwanza, kisha outer loop husasisha target path length kwa binary search. Inner genetic algorithm huzalisha candidate baffle layouts kwa hot na cold sides kwa kila target; katika final target length, five best designs huthibitishwa kwa full CFD na final layout huchaguliwa.

Representative flow path ilitolewaje?

Katika graph ya accessible cells kwa kila channel, shortest path kati ya inlet na outlet iliamuliwa. Ikiwa kulikuwa na multiple paths zenye same length, path yenye turns chache zaidi ilichaguliwa:

\[ P^Q=\left(p^Q_1,p^Q_2,\ldots,p^Q_{n_Q}\right), \qquad p^Q_1=s^Q,\quad p^Q_{n_Q}=t^Q \]

Hapa \(P^Q\) ni representative path; \(p^Q_m\) ni center ya cell kwenye path; \(s^Q\) na \(t^Q\) ni inlet na outlet nodes. Path length kwa idadi ya grid edges ni:

\[ L(P^Q)=n_Q-1 \]

Kabla ya path kutolewa, redundant wall segments zisizobadilisha inlet-outlet connectivity au local width ziliondolewa. Hatua hii inalenga kuzuia genetic algorithm kutathmini kama advantage baffles ambazo haziathiri physical flow.

Flow priors zinawakilisha matatizo gani ya kimwili?

1. Ulinganifu na target path length

Ukaribu wa representative path na target kutoka outer loop hupimwa kwa penalty:

\[ f^Q_{\mathrm{uzunluk}}=\left|L(P^Q)-L_{\mathrm{tar}}\right| \]

Path ikiwa fupi sana, flow inaweza kutotumia sehemu ya kutosha ya core. Path ikiwa ndefu kupita kiasi, inaweza kuongeza pressure drop. Term hii inalenga kuhakikisha kwamba baffles zinaunda route inayolingana na hydraulic budget.

2. Inlet-weighted turn penalty

Kila mabadiliko ya direction katika flow path yanaweza kuunda local loss. Katika utafiti, turns zilizo karibu na inlet zimepewa weight kubwa zaidi:

\[ w_{\mathrm{konum}}(i)=1-\frac{i}{n_Q} \]

\(i\) ni nafasi ya turn kwenye path; \(n_Q\) ni total node count ya path. Kwa kuwa turn karibu na inlet inaweza kuathiri entire downstream region, hupewa penalty kubwa kuliko turn karibu na outlet. Sharp turns zilizo karibu sana pia hupata penalty kubwa kuliko isolated turn baada ya long straight section.

3. Channel-width regularity

Local channel widths zilipimwa kando ya representative path na coefficient of variation ikakokotolewa:

\[ CV_w^Q=\frac{\operatorname{std}\left(\{w_m^Q\}\right)}{\operatorname{mean}\left(\{w_m^Q\}\right)} \]

\(w_m^Q\) ni accessible channel width iliyopimwa perpendicular to flow direction katika sehemu husika ya path. Frequent narrowing na widening zinaweza kuvuruga flow distribution na kuunda local pressure losses. Small coefficient of variation inawakilisha channel structure iliyo regular zaidi.

4. Dead-region na unnecessary-side-route penalty

Kwa kila accessible cell, jumla ya shortest distances kutoka inlet hadi cell na kutoka cell hadi outlet ilikokotolewa. Kadiri jumla hii ilivyo ndefu kuliko direct shortest inlet-outlet path, cell hiyo ilitathminiwa zaidi kama inefficient side route:

\[ s^Q_{r,c}=d^Q_{\mathrm{giriş},r,c}+d^Q_{\mathrm{çıkış},r,c}-L^Q_{\mathrm{kısa}} \]

Cells zilizo kwenye shortest route zina excess value ya sifuri. Cells zinazoweza kufikiwa lakini zinahitaji long detour hupata positive penalty. Disconnected cells hupewa heavy penalty yenye ukubwa wa mara kumi ya shortest-path length. Channel-level dead-region metric ni wastani wa accessible cells zote.

5. Hot-cold counterflow matching

Katika ideal counterflow, eneo karibu na hot-flow inlet linapaswa kulingana na cold-flow outlet, na eneo karibu na hot-flow outlet na cold-flow inlet. Utafiti unalenga kufanya jumla ya normalized inlet distances katika channels mbili iwe karibu na moja:

\[ \frac{d^c_{r,c}}{L_c}+\frac{d^h_{r,c}}{L_h}\approx1 \]

Deviation kutoka matching hii hupimwa kwenye all commonly accessible cells na kubadilishwa kuwa counterflow penalty \(f_{\mathrm{cf}}\). Kwa njia hii, method haipangi cold na hot channels kila moja peke yake tu; pia inatathmini kama two flows hukutana katika thermally suitable locations kando ya common TPMS wall.

Combined prior objective

Kwa single channel, path length, turn, width na dead-zone metrics zimeunganishwa kwa weights:

\[ f^Q_{\mathrm{tek}}=\lambda_{\mathrm{uzunluk}}\frac{f^Q_{\mathrm{uzunluk}}}{L^Q_{\mathrm{kısa}}}+\lambda_{\mathrm{dönüş}}f^Q_{\mathrm{dönüş}}+\lambda_{\mathrm{genişlik}}f^Q_{\mathrm{genişlik}}+\lambda_{\mathrm{ölü}}\frac{f^Q_{\mathrm{ölü}}}{L^Q_{\mathrm{kısa}}} \]

Joint objective function ya cold na hot channels ni:

\[ F(B^c,B^h;L_{\mathrm{tar}})=f^c_{\mathrm{tek}}+f^h_{\mathrm{tek}}+\lambda_{\mathrm{cf}}f_{\mathrm{cf}} \]

\(\lambda\) coefficients huamua relative importance ya priors tofauti. Hata hivyo, numerical values za coefficients hizi hazijaripotiwa katika utafiti. Upungufu huu hufanya iwe vigumu kuendesha algorithm upya kwa independently with same settings.

Zaidi ya hayo, equations zinajenga \(F\) kama jumla ya penalty terms na zinaeleza kwamba lower values ni better, lakini katika sehemu ya alternating genetic algorithm imeandikwa kwamba \(F\) “maximized”. Ikiwa hakukutumiwa unreported fitness transformation, kauli hii ni methodological inconsistency kuhusu objective direction.

Genetic algorithm iliendeshwaje?

Kuboreshwa kwa wakati mmoja kwa binary baffle variables zote kwenye cold na hot sides huunda high-dimensional combinatorial problem. Kwa hiyo watafiti walitumia alternating optimization:

1. Hot side iliwekwa fixed huku baffles za cold side zikiboreshwa.
2. Cold side iliyosasishwa iliwekwa fixed na baffles za hot side zikaboreshwa.
3. Mchakato huu ulirudiwa hadi two sides zilingane kupitia counterflow matching.

Katika kila single-side genetic-algorithm stage, population ya 100 individuals na 200 generations ilitumika. Parents walichaguliwa kwa tournament selection, crossover ilitumika, na kila binary wall variable ilibadilishwa independently kwa mutation probability ya 0,05. Best individual ilihifadhiwa kwa elitism.

Baada ya crossover na mutation, candidates zilipitia repair process. Process hii iliondoa disconnected au floating obstacle layouts, kurejesha inlet-outlet connectivity na kusogeza representative path length karibu na target value ya outer loop. Candidates zilizorekebishwa na zinazotimiza connectivity conditions pekee ndizo zilipata alama kwa prior objective.

CFD model na boundary conditions

Full flow na heat-transfer validation ilifanywa kwa COMSOL Multiphysics 6.4. Model inatatua conjugate heat transfer kati ya stainless-steel TPMS solid na two counterflowing water circuits.

• Flow ilichukuliwa kuwa steady na incompressible.
• Navier-Stokes na energy equations zilituliwa pamoja.
• Cold-flow inlet temperature iliwekwa 293 K, hot-flow inlet temperature 353 K.
• Velocity boundary condition ilitumika kwenye inlets, pressure boundary condition kwenye outlets.
• No-slip, temperature continuity na heat-flux continuity zilitumika kwenye fluid-solid interfaces.
• Outer walls zilichukuliwa kuwa thermally insulated.
• Methods zote zililinganishwa kwa same mesh, boundary conditions, solver settings na post-processing method.

Mesh-independence check

Mesh sensitivity ilichunguzwa katika final baffled design ya G-type, speed 2 m/s, depth 1 na target 60 kPa. COMSOL Normal, Extra fine na Extremely fine meshes zililinganishwa.

Tofauti ya heat transfer kati ya Extra fine na Extremely fine meshes ilikuwa %0,18 pekee, huku total pressure-drop difference kwa pande mbili ikiwa takribani %1,05. Watafiti walitumia Extra fine mesh katika all subsequent analyses ili kuepuka higher computational cost.

Representative 20 kPa comparison ilionyesha nini?

Controlled comparison ilifanywa kwa P-type, one-period, inlet velocity 1 m/s na pressure-drop limit 20 kPa kwa pande zote mbili. Baffle-free reference, direct genetic algorithm na flow-prior genetic algorithm zililinganishwa chini ya same CFD solution conditions.

Baffle-free core ilifikia 15,69 kPa pressure drop kwenye hot side huku ikitumia 0,88 kPa pekee kwenye cold side. Best feasible design ya direct genetic algorithm pia haikutumia sehemu kubwa ya hydraulic allowance kwenye cold side, ikiwa na 1,98 kPa. Proposed design ilileta flow circuits zote mbili karibu na 20 kPa limit na kubadilisha available hydraulic allowance kuwa thermal capacity kwa uwiano bora zaidi.

Figure 3 inaonyesha kwamba direct genetic algorithm iliacha route fupi na rahisi kiasi kwenye cold side, huku flow-prior method ikizungusha cold flow katika sehemu kubwa zaidi ya core. Path kwenye hot side iliwekwa rahisi zaidi, na different representative lengths zilichaguliwa kwa channels mbili. Final cold/hot path length ni 57/15.

Velocity fields katika Figure 4 zinaonyesha kwamba flow-prior design ilisambaza high-speed regions kwenye sehemu kubwa zaidi ya core. Katika baffle-free structure, velocity hukusanyika kwenye corridors fulani, wakati katika proposed structure cold na hot streamlines husweep area kubwa zaidi ya TPMS. Temperature map pia inaonyesha thermal transition zone kati ya two flows ikiwa pana na regular zaidi.

Tathmini kumi za CFD zilitumikaje?

Tathmini 10 za proposed method ziligawanywa katika hatua mbili. First five evaluations zilitumika kwa outer loop kutafuta path length inayolingana na pressure limit. Next five evaluations zilitumika kuthibitisha kwa full CFD top five candidates za genetic algorithm katika fixed 57/15 path length.

Baadhi ya candidates katika pressure-search stage zilizidi 20 kPa limit. Hii inaonyesha kwamba si intermediate designs zote za method zilikuwa feasible. Kazi ya outer loop ni kutumia violations hizi kurekebisha path-length bound na kusogeza final candidate set karibu na pressure limit.

Best feasible result ya first stage ilikuwa 36,14 W; baada ya validation ya final five candidates, 37,96 W ilifikiwa. Kwa hiyo hata first stage pekee ilizidi baffle-free na direct-GA references; second stage ilitoa additional improvement ya 1,82 W.

Matokeo katika hali sita za uendeshaji

Highest improvement ya %78,6 ilipatikana katika P-type 25 kPa case, huku lowest improvement ya %24,4 ikipatikana katika longer 7×12 G-type design area. Watafiti wanasema kwamba katika long na narrow structures, flow inaweza kuwa regular zaidi na developed zaidi katika through-flow direction hata bila baffle; kwa hiyo stagnant area inayoweza kuanzishwa upya hubaki kidogo. Wakati huo huo, narrow geometry huongeza pressure penalty ya new baffles kwa haraka zaidi.

Katika conditions zenye pronounced dead zones zaidi au imbalanced flow distribution, improvement room ya baffles ilikuwa kubwa zaidi. Hata hivyo, six conditions zina inlet velocity, TPMS structure, geometric depth na pressure limit tofauti. Kwa hiyo absolute values kama 113,13 W na 37,96 W hazipaswi kutumiwa kwa direct performance ranking kati ya conditions.

Pressure-drop graphs zinaonyesha nini?

Figure 6 inaonyesha pressure drops kwenye cold na hot sides wakati wa outer search na final five candidates. Katika P-type cases, symmetric path lengths zilipochaguliwa kwa cold na hot sides, hot-side pressure drop iliongezeka sana huku cold side ikibaki chini ya hydraulic limit. Outer loop ilirekebisha imbalance hii kwa kuchagua different \(L_c/L_h\) pairs.

Katika high-speed 60 kPa G-type case, shortest path pair pekee katika first stage ndiyo ilitimiza two-sided pressure limit. Kati ya final five validation candidates, first, second na fifth designs pekee ndizo zilibaki feasible. Matokeo haya yanathibitisha kwamba representative path length ni useful predictor ya pressure drop lakini si perfect.

Heat-transfer graphs zinapaswa kutafsiriwaje?

Katika Figure 7, green regions zinaonyesha candidates zinazotimiza pressure limit, na gray horizontal lines zinaonyesha baffle-free references. Final design haikuchaguliwa kwa highest \(Q\) pekee, bali kwa sharti la kwanza la kutimiza pressure limit.

Baadhi ya candidates zilizozidi pressure limit ziliweza kutoa higher au similar thermal results; lakini hazikukubaliwa kama feasible designs. Tofauti hii ni muhimu: kuongeza heat transfer kwa kuongeza baffles pekee si mafanikio. Increase lazima ipatikane ndani ya defined hydraulic allowance.

Ablation study ilionyesha priors zipi ni muhimu?

Watafiti waliweka path-length control fixed na kuondoa other four priors moja baada ya nyingine. Variants zote zilitathminiwa kwenye 57/15 path pair.

Width regularity ilipoondolewa, heat transfer ilibaki karibu na full method, lakini hot-side pressure drop ilifikia 66,57 kPa, zaidi ya mara tatu ya limit. Term hii huzuia excessive hydraulic resistance kukusanyika katika narrow bottlenecks chache.

Inlet-weighted turn penalty ilipoondolewa, limit ilizidiwa kwa pande zote mbili. Matokeo haya yanaonyesha kwamba especially near-inlet na frequent turns ni muhimu katika kudhibiti local pressure loss.

Dead-zone penalty ilipoondolewa, heat transfer ilipungua kwa 0,24 W pekee; hot side ilizidi limit kwa 0,72 kPa. Kwa kuwa baseline layout iliyochaguliwa tayari ilikuwa na dead zones chache, thermal effect ya term hii ilikuwa ndogo, huku hydraulic-regulating effect ikiwa dhahiri zaidi.

Hot-cold counterflow matching ilipoondolewa, pressure limit ilidumishwa, lakini \(Q\) ilipungua kutoka 37,96 W hadi 36,20 W. Term hii haifanyi kazi kwa hydraulic feasibility, bali kwa kuboresha thermal matching ya two flows kwenye common surface.

Mechanism kuu iliyoonyeshwa na utafiti

Matokeo yanaonyesha kwamba kuongeza vizuizi zaidi kwenye TPMS core pekee hakutengenezi heat exchanger bora moja kwa moja. Location ya baffles, paths za hot na cold flows, local channel width na distribution ya pressure drop kati ya two circuits ni muhimu kwa pamoja.

Mafanikio ya proposed method yanatokana na three linked effects:

1. Flow ilielekezwa mbali na low-resistance short-circuit paths kati ya inlet na outlet.
2. Sehemu kubwa zaidi ya TPMS core ilisweepiwa na high au medium velocity flow.
3. Directions za hot na cold flows kando ya common wall zililetwa kwenye counterflow arrangement inayofaa zaidi.

Hivyo, chini ya same upper pressure-drop limit, internal surface kubwa zaidi ilishiriki katika useful heat transfer. Dhana ambayo watafiti wanaita “surface activation” ni kutumia kwa kweli available geometric area kupitia flow na temperature difference.

Utafiti unaunga mkono nini?

Utafiti unaunga mkono kwamba chini ya specific numerical conditions, flow-prior-guided baffle placement inaweza kuzalisha heat transfer kubwa zaidi katika TPMS cores kuliko baffle-free design bila kuzidi pressure-drop limit.

Katika representative 20 kPa case, same binary wall representation, repair method, CFD solution na pressure limit zilipotumika, flow-prior method ilifikia \(Q\) kubwa zaidi kuliko direct genetic algorithm kwa takribani one-tenth CFD evaluations.

Matokeo katika six conditions yanaonyesha kwamba method iliweza kupata feasible candidates katika different P na G structures, velocities, geometric depths na pressure limits. Hata hivyo, kwa kuwa direct genetic-algorithm comparison ilikamilishwa katika condition moja tu, algorithmic superiority katika all conditions haijathibitishwa.

Utafiti hauthibitishi nini?

• Haijaonyeshwa kwamba optimized baffles zinaweza kutengenezwa bila defects kwa metal additive manufacturing.
• Heat transfer au pressure drop haijapimwa kwenye real prototype.
• Surface roughness, printing defects, channel blockage, fouling na long-term corrosion effects hazikumodeliwa.
• Athari ya baffles kwenye mechanical strength na fatigue life ya TPMS core haijachunguzwa.
• Total pumping power haijaboreshwa kama direct objective function; upper pressure-drop limit imetumika kama approximate hydraulic budget.
• Variable-flow, transient au two-phase operating conditions hazijatathminiwa.
• Haijaonyeshwa kwamba proposed method ni bora katika all TPMS topologies au all Reynolds numbers.
• Genetic weights na baadhi ya reproducibility settings za algorithm hazijaripotiwa kikamilifu.
• Method haijapitia peer review.

Thamani ya kihandisi kwa Türkiye

Nchini Türkiye, kuna mahitaji ya heat exchangers zinazotoa heat transfer kubwa zaidi katika volume ndogo katika energy, defense, aerospace, hydrogen systems, heat pumps, power electronics, battery thermal management na waste-heat recovery. Kwa kuwa TPMS structures zinaweza kutengenezwa kwa metal additive manufacturing, utafiti huu hautoi theoretical geometry analysis pekee, bali pia preliminary-design method inayoweza kuunganishwa na advanced manufacturing.

Vipengele vya method vinavyoweza kuhamishwa kwenye R&D na product development nchini Türkiye ni:

• Kuboreshwa kwa baffle placement kabla ya printing kwa locally manufactured TPMS samples,
• Kutumia upper pressure-drop limit moja kwa moja kama design constraint katika systems zenye limited pump au compressor power,
• Kutengeneza geometry-specific flow priors ili kupunguza CFD cost,
• Kuboreshwa kwa hot na cold flow channels kwa mutual thermal matching badala ya independently,
• Kuongeza additive-manufacturing tolerance, minimum wall thickness na support requirement kwenye repair stage ya genetic algorithm,
• Kuthibitisha numerical results kwa majaribio kwa pressure sensors, flow meters na calorimetric heat balance.

Katika real Turkish application, utafiti haupaswi kutumiwa kama final design recipe moja kwa moja. Ikiwa badala ya water itatumika oil, air, refrigerant au fuel; au temperature range, material, surface quality na pump efficiency zitabadilika, optimization inapaswa kufanywa upya.

Mbinu na Matokeo ya Utafiti

Muhtasari wa kiufundi wa mbinu

Matokeo makuu ya representative comparison

• Proposed method ilitoa 37,96 W, heat transfer ya %75,5 juu ya baffle-free reference.
• Ongezeko dhidi ya 28,97 W ya direct genetic algorithm ni takribani %31.
• Proposed method ilitumia 10 CFD evaluations, direct GA takribani 100.
• Cold na hot side effectiveness iliongezeka hadi %82,54.
• Pressure drops zilikuwa 19,38 na 19,79 kPa, chini ya 20 kPa limit.
• Direct GA ilitumia 1,98 kPa pekee kwenye cold side na kuacha hydraulic allowance ikiwa imbalanced.
• Final representative path lengths ni 57 grid steps kwenye cold side na 15 kwenye hot side.

Common findings kutoka six conditions

• Baffled designs zilitoa \(Q\) kubwa kuliko baffle-free reference katika all conditions.
• Feasible gain range ni %24,4–%78,6.
• Improvement ratio ilipungua katika longer na narrower design area.
• Katika cores zenye imbalanced flow distribution zaidi, surface area inayoweza kuanzishwa upya na baffles ilikuwa kubwa zaidi.
• Representative path length ilihusiana na pressure drop lakini haikutoa exact feasibility guarantee peke yake.
• CFD validation ya final five candidates iliboresha first-stage result kwa kiasi kidogo lakini kinachopimika katika baadhi ya conditions.

Nguvu za utafiti

• Controlled algorithm comparison ilifanywa chini ya same binary design representation na same CFD protocol.
• Pressure drop ilitumika kama direct feasibility constraint, si output inayoripotiwa baadaye tu.
• Heat-transfer increase ilichunguzwa katika mechanism level kwa velocity, temperature na streamlines za two sides.
• Numerical-solution sensitivity ilikaguliwa kwa mesh-independence study.
• Different TPMS structures, velocities, geometric depths na pressure limits zilijaribiwa.
• Ablation analysis ilitathmini separately hydraulic na thermal role ya kila physical prior.
• CFD kutotumika katika every genetic-algorithm step kulipunguza computational cost kwa kiasi kikubwa.

Mapungufu ya reproducibility na interpretation

• Numerical values za \(\lambda\) weights zinazounganisha flow priors hazijaripotiwa.
• Ingawa penalty function inapaswa minimized katika equations, algorithm description inatumia neno “maximized”.
• Stopping criterion ya genetic algorithm na exact number of alternating rounds hazijaelezwa kwa kina.
• All physical dimensions za TPMS na design domains hazijawasilishwa kwa undani wa kutosha katika main text.
• Direct GA comparison ipo katika one of six conditions pekee.
• Ingawa CFD evaluation counts zimelinganishwa, total CPU time na hardware information hazijatolewa.
• Hakuna experimental prototype validation.
• Temperature-dependent variation ya material properties haijaripotiwa kwa undani.
• Manufacturing tolerance, minimum printable channel, support structures na powder-removal requirements hazijaingizwa katika optimization.
• Structural stress, thermal fatigue na pressure resistance hazijachunguzwa.
• Fouling na long-term operation effects hazikumodeliwa.

Dokezo la Chanzo na Mbinu

• Jina kamili la asili la utafiti: Baffle optimization for TPMS compact heat exchangers via flow priors under hydraulic constraints
• Waandishi na mpangilio: Lingxin Cao; Xiangjun Wu; Hanyue Xu; Xiaowei Duan; Bingteng Sun; Qiang Du; Dechun Li; Lin Lu
• Equal contribution au equal first authorship: Haijatajwa.
• Corresponding author: Lin Lu
• Corresponding author email: Haijatajwa katika utafiti.
• Institution 1: School of Information Science and Engineering, Shandong University, Qingdao, Shandong, China
• Institution 2: School of Computer Science and Technology, Shandong University, Qingdao, Shandong, China
• Institution 3: Advanced Gas Turbine Laboratory, Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing, China
• DOI: 10.2139/ssrn.6944970
• Publication platform: SSRN
• Platform operator: Elsevier
• Original publisher: Haitumiki. SSRN ni preprint platform na version iliyochunguzwa si peer-reviewed journal publication.
• Peer-reviewed journal: Hakuna accepted au published peer-reviewed journal name iliyotajwa katika utafiti.
• Publication year: 2026
• Source type: Preprint ya CFD-based numerical optimization research
• Peer-review status: Utafiti haujapitia peer review.
• Official SSRN link:SSRN study page
• DOI link:10.2139/ssrn.6944970

Uploaded version inatumia SSRN record 6944970 kama DOI na footer. Pia kuna earlier SSRN record 6878424 yenye same original title na author list. Utafiti hauelezi kama relation kati ya records hizi mbili ni previous version, corrected version au duplicate record. Kwa hiyo citation inapaswa kutumia 10.2139/ssrn.6944970, inayolingana na current examined file.

Research iliungwa mkono na National Natural Science Foundation of China kupitia grants U25A20438, 62472258 na 52488101, na pia na Qingdao Postdoctoral Science Foundation kupitia grant QDBSH20250202009.

Author contribution roles, data au code access link na explicit conflict-of-interest statement hazipo katika utafiti. Genetic-algorithm implementation code, COMSOL model files na optimized binary wall matrices zote hazijawasilishwa kwa open archive link.

Maelezo haya ya Kiswahili yameandaliwa kwa kuchunguza text, equations, algorithm flowchart, baffle layouts, velocity na temperature fields, streamlines, pressure-drop graphs, heat-transfer graphs, tables na ablation comparisons. Scientific content inategemea examined study pekee; external sources zimetumika tu kwa bibliographic verification ya DOI, author record, platform na publication status.

Matokeo ya utafiti yako katika numerical-design level. Reported increases za %24,4–%78,6 si experimental device measurements, bali ni results za COMSOL simulations chini ya geometry, fluid, velocity, temperature na pressure-drop limits maalum. Ili findings zibadilishwe kuwa real product performance, additive-manufactured prototype, pressure na temperature measurements, calorimetric validation, manufacturing-tolerance analysis na long-term durability tests zinahitajika.