Evaluation of the performance of a product and its components under impact loading is one of the key considerations in design. In order to assess resistance to damage or ability to absorb energy through plastic deformation of a structural component, impact testing is often carried out to obtain the 'Force - Displacement' response of the deformed component. In this context, it may be noted that load cells and accelerometers are commonly used as sensors for capturing impact responses. A drop-weight impact testing set-up consisting of a moving impactor head with a lightweight piezoresistive accelerometer and a strain gage based compression load cell mounted on it is used to carry out the impact tests. The basic objective of the present study is to assess the accuracy of responses recorded by the said transducers, when these are mounted on a moving impactor head. In the present work, a novel approach of theoretically evaluating the responses obtained from this drop-weight impact testing set-up for different axially loaded specimen has been executed with the formulation of an equivalent lumped parameter model (LPM) of the test set-up. For the most common configuration of a moving impactor head mounted load cell system in which dynamic load is transferred from the impactor head to the load cell, a quantitative assessment is made of the possible discrepancy that can result in load cell response. Initially, a 3-DOF (degrees-of-freedom) LPM is considered to represent a given impact testing set-up with the test specimen represented with a nonlinear spring. Both the load cell and the accelerometer are represented with linear springs, while the impacting unit comprising an impactor head (hammer) and a main body with the load cell in between are modelled as rigid masses. An experimentally obtained force-displacement response is assumed to be a nearly true behaviour of a specimen. By specifying an impact velocity to the rigid masses as an initial condition, numerical solution of the governing differential equations is obtained using Implicit (Newmark-beta) and Explicit (Central difference) time integration techniques. It can be seen that the model accurately reproduces the input load-displacement behaviour of the nonlinear spring corresponding to the tested component, ensuring the accuracy of these numerical methods. The nonlinear spring representing the test specimen is approximated in a piecewise linear manner and the solution strategy adopted and implemented in the form of a MATLAB script is shown to yield excellent reproduction of the assumed load-displacement behaviour of the test specimen. This prediction also establishes the accuracy of the numerical approach employed in solving the LPM system. However, the spring representing the load cell yields a response that qualitatively matches the assumed input load-displacement response of the test specimen with a lower magnitude of peak load. The accelerometer, it appears, may be capable of predicting more closely the load experienced by a specimen provided an appropriate mass of the impactor system i.e. impacting unit, is chosen as the multiplier for the acceleration response. Error between input and computed (simulated) responses is quantified in terms of root mean square error (RMSE). The present study additionally throws light on the dependence of time step of integration on numerical results. For obtaining consistent results, estimation of critical time step (increment) is crucial in conditionally stable central difference method. The effect of the parameters of the impact testing set-up on the accuracy of the predicted responses has been studied for different combinations of main impactor mass and load cell stiffness. It has been found that the load cell response is oscillatory in nature which points out to the need for suitable filtering for obtaining the necessary smooth variation of axial impact load with respect to time as well as deformation. Accelerometer response also shows undulations which can similarly be observed in the experimental results as well. An appropriate standard SAE-J211 filter which is a low-pass Butterworth filter has been used to remove oscillations from the computed responses. A load cell is quite capable of predicting the nature of transient response of an impacted specimen when it is part of the impacting unit, but it may substantially under-predict the magnitudes of peak loads. All the above mentioned analysis for a 3 DOF model have been performed for thin-walled tubular specimens made of mild steel (hat-section), an aluminium alloy (square cross-section) and a glass fibre-reinforced composite (circular cross-section), thus confirming the generality of the inferences drawn on the computed responses. Further, results obtained using explicit and implicit methodologies are compared for three specimens, to find the effect, if any, on numerical solution procedure on the conclusions drawn. The present study has been further used for investigating the effects of input parameters (i.e. stiffness and mass of the system components, and impact velocity) on the computed results of transducers. Such an investigation can be beneficial in designing an impact testing set-up as well as transducers for recording impact responses. Next, the previous 3 DOF model representing the impact testing set-up has been extended to a 5 DOF model to show that additional refinement of the original 3 DOF model does not substantially alter the inferences drawn based on it. In the end, oscillations observed in computed load cell responses are analysed by computing natural frequencies for the 3 DOF lumped parameter model. To conclude the present study, a 2 DOF LPM of the given impact testing set-up with no load cell has been investigated and the frequency of oscillations in the accelerometer response is seen to increase corresponding to the mounting resonance frequency of the accelerometer. In order to explore the merits of alternative impact testing set-ups, LPMs have been formulated to idealize test configurations in which the load cell is arranged to come into direct contact with the specimen under impact, although the accelerometer is still mounted on the moving impactor head. One such arrangement is to have the load cell mounted stationary on the base under the specimen and another is to mount the load cell on the moving impactor head such that the load cell directly impacts the specimen. It is once again observed that both these models accurately reproduce the input load-displacement behaviour of the nonlinear spring corresponding to the tested component confirming the validity of the model. In contrast to the previous set-up which included a moving load cell not coming into contact with the specimen, the spring representing the load cell in these present cases yields a response that more closely matches the assumed input load-displacement response of a test specimen suggesting that the load cell coming into direct contact with the specimen can result in a more reliable measurement of the actual dynamic response. However, in practice, direct contact of the load cell with the specimen under impact loading is likely to damage the transducer, and hence needs to be mounted on the moving head, resulting in a loss of accuracy, which can be theoretically estimated and corrected by the methodology investigated in this work.

Optimizing and controlling the heating systems in buildings is one way to decrease their load on the power grid, as well as introduce load flexibility to be used in Demand Response (DR) applications. A requirement in occupied buildings is that the thermal comfort of the residents is guaranteed, making the optimization of heating systems a constrained problem with respect to indoor temperature. Thermal models capable of predicting indoor temperatures over short (24 hour) horizons are one way to guarantee this comfort. The accuracy and computational complexity of these models have the most significant impact on controller performance. The data requirements and the expert knowledge required for model implementation are also important factors, since they determine the development costs and, finally, whether a model is feasible to implement. First a literature study explores current modeling approaches that depend only on time-series sensor data and that are suited for control applications. A modeling type found to be fit for such purposes are grey-box models, specifically physically inspired inverse models whose parameters are estimated based on data, such as Resistance- Capacitance (RC) models. This modeling of a dynamical system approach uses prior information in the form of the assumed physical equations and has the potential to increase the performance on sparse data problems. The simple form of the model also has a low level of complexity, making it well suited for control applications. However, expert knowledge can be needed for choosing the model equations as well as initializing the parameters. Then the effects of varying RC model complexity, parameter initialization, and training data are investigated in the case study. The chosen models are 1R1C, 2R2C, and 3R2C. They are fitted using the Nelder-Mead algorithm and validated using the data collected from the RISE Research Villa. Parameter initializations are varied by two orders of magnitude and then fitted on different data sequences to avoid relying on expert knowledge in model creation. The initializations that converged with the best R2 training fit on all sequences were deemed reasonable initializations for the problem and used in the training length comparison. The training length of the models varies from 24 to 384 hours. The results showed that increased training data length correlates positively with performance up to 192 hours for all models, but further increasing it gave inconclusive results. The higher order models evaluated struggled to beat the simplest model or even the constant prediction baseline in Mean Absolute Error (MAE) performance at all training lengths, indicating either that the models selected are unsuitable or that the data features chosen are unrepresentative of the indoor temperature dynamics. Regardless, the MAE errors presented here are comparable to the outcomes of related works. This is possibly an artifact of this dataset having a low variance in temperature and thus resulting in lower errors, which underlines the importance of the data used in case-studies. / Att optimera och styra värmesystemen i byggnader är ett sätt att minska belastningen på elnätet och införa flexibilitet i belastningen som kan användas i tillämpningar för efterfrågeflexibilitet (Demand Response, DR). Ett krav i bebodda byggnader är att de boendes termiska komfort garanteras, vilket gör optimeringen av värmesystemen till ett begränsat problem med avseende på inomhustemperaturen. Termiska modeller som kan förutsäga inomhustemperaturer på kort sikt (24 timmar) är ett sätt att garantera denna komfort. Dessa modellers noggrannhet och beräkningskomplexitet har störst inverkan på styrningens prestanda. Datakraven och den expertkunskap som krävs för att genomföra modellen är också viktiga faktorer, eftersom de avgör utvecklingskostnaderna och slutligen om en modell är möjlig att implementera. Först görs en litteraturstudie av nuvarande modelleringsmetoder som endast är beroende av tidsserier av sensordata och som lämpar sig för reglertillämpningar. En modelleringstyp som visat sig vara lämplig för sådana ändamål är grey-box-modeller, särskilt fysikaliskt inspirerade inversa modeller vars parametrar estimeras på basis av data, t.ex. RC-modeller (Resistance-Capacitance). Denna modell av ett dynamiskt system modellering använder förhandsinformation i form av de antagna fysiska ekvationerna och har potential att öka prestandan vid problem med begränsad data. Modellens enkla form har också en låg komplexitetsnivå, vilket gör den väl lämpad för kontrolltillämpningar. Expertkunskap kan dock behövas för att välja modellekvationer och initiera parametrarna. Därefter undersöks effekterna av att variera RC-modellens komplexitet, parameterinitialisering och träningsdata i fallstudien. De valda modellerna är 1R1C, 2R2C och 3R2C. De tränas med hjälp av Nelder-Mead-algoritmen och valideras med hjälp av data som samlats in från RISE Research Villa. Initialiseringarna av parametrarna varieras med två storleksordningar och anpassas sedan på olika dataserier för att undvika att förlita sig på expertkunskap vid skapandet av modellerna. De initialiseringar som konvergerade med den bästa träningsanpassningen R2 på alla sekvenser ansågs vara rimliga initialiseringar för problemet och användes i jämförelsen av träningslängden. Modellernas träningslängd varierar mellan 24 och 384 timmar. Resultaten visade att en ökad längd på träningsdata korrelerar positivt med prestanda upp till 192 timmar för alla modeller, men att ytterligare ökning inte ger några entydiga resultat. De utvärderade modellerna av högre ordning hade svårt att överträffa den enklaste modellen eller till och med referensmodellen med konstant prediktion i fråga om genomsnittligt absolut fel (MAE) vid alla träningslängder, vilket tyder antingen på att de valda modellerna är olämpliga eller att de valda datafunktionerna inte är representativa för inomhustemperaturens dynamik. Oavsett detta är de MAE-fel som presenteras här jämförbara med resultaten från relaterade studier. Detta är möjligen en artefakt av att detta dataset har en låg varians i temperaturen och därmed resulterar i lägre fel, vilket understryker vikten av de data som används i fallstudier.

A soft quadruped robot is being developed at the Department of Machine Design and Department of Production Engineering at KTH. The legs of the robot consist of four continuum actuators that can achieve complex movements. In order to efficiently develop gaits for the robot, reinforcement learning will be used. The learning process will use data from simulation instead of directly from the real robot to save time and resources. However, it is significantly more computationally expensive to simulate soft robotics than rigid, because the physical laws of flexible materials are inherently complex. Because of this, soft robot simulations tend to be slower which limits their usability for reinforcement learning. This thesis explores simulation modelling options in Matlab Simscape for the soft quadruped robot, that can be used in reinforcement learning. Four simulation models of the soft actuator were implemented in order to be tested and compared. Two actuation methods and two build options were chosen based on the literature study and related works, and were then permuted for the different combinations. The tested combinations are: lumped-parameter method actuated by internal force, flexible beam actuated by internal force, lumped-parameter method actuated by cable/pulley network and flexible beam actuated by cable/pulley network. The four actuators were built and tested separately. Computational time and simulation-to-reality gap were used for evaluating the modeling methods. The results show that the best option when modelling the soft actuator for reinforcement learning in Matlab Simscape is to use the lumped-parameter method in combination with a cable and pulley network. High accuracy level can still be achieved despite not keeping the true number of attachment points between the cable and actuator. The number of pulleys in the model is linearly correlated to the time cost required to simulate the model. / En mjuk fyrbent robot är under utveckling vid institutionen för maskinkonstruktion och institutionen för industriell produktion på KTH. Robotens ben består av fyra kontinuerligt deformerbara ställdon som kan åstadkomma komplexa rörelser. För att effektivt utveckla gångstilar till roboten kommer förstärkt inlärning att användas. Inlärningsprocessen kommer att använda data från simulering istället för från den fysiska roboten för att spara tid och resurser. Det är dock betydligt dyrare beräkningsmässigt att simulera mjuk robotik än styv, eftersom flexibla material är mer komplexa. På grund av detta tenderar simuleringar av mjuka robotar att vara långsammare, vilket begränsar deras användbarhet för förstärkt inlärning. Detta examensarbete utforskar därför alternativ för modellering och simulering av den mjuka fyrbenta roboten i Matlab Simscape, med målet att den ska kunna användas med förstärkt inlärning. Fyra olika simuleringsmodeller av det mjuka ställdonet implementerades för att testas och jämföras. Två aktiveringsmetoder och två konstruktionsalternativ valdes baserat på litteraturstudien och relaterade arbeten, och permuterades sedan till möjliga versioner. De testade versionerna är således: klumpparametermetod som aktiveras av intern kraft, flexibel balk som aktiveras av intern kraft, klumpparametermetod som aktiveras av kabelnätverk och flexibel balk som aktiveras av kabelnätverk. De fyra ställdonen byggdes och testades separat. Beräkningstid och grad av verklighetstrogenhet, användes för att jämföra resultaten av dessa tester. Resultaten visar att det bästa alternativet vid modellering av det mjuka ställdonet för förstärkt inlärning i Matlab Simscape är att använda klumpparametermetoden i kombination med ett kabelnätverk. Hög noggrannhetsnivå kan uppnås trots att man inte bibehåller det verkliga antalet fästpunkter mellan kabeln och ställdonet. Antalet fästpunkter för kabeln i modellen är linjärt korrelerat till den tidskostnad som krävs för att simulera modellen.

A significant challenge with permanent magnet synchronous motors (PMSMs) is thermal management. Thermal stress can lead to irreversible damage to components, and to enable efficient cooling, a thermal model is needed. In this thesis paper, methods for estimating the hot spot temperature of the windings in PMSMs used in heavy-duty EVs are investigated. The methods include black-box models and lumped parameter thermal network-based models. The results reveal that the implemented models are not sufficient for achieving the desired accuracy, and indicate that more parts of the windings need to be considered.

No description available.

In general, electric propulsion offers very high efficiency but relatively low thrust. To remedy this, several ion engines can be assembled in a clustered configuration and operated in parallel. This requires the careful design of a frame to accommodate the individual propulsion systems. This frame must be modular to be used in different cluster sizes, and verify thermal and mechanical requirements to ensure the nominal operation of the thrusters. The present report aims to show the design process of such a frame, from preliminary modelling to the experimental study of a prototype. This document features an overview of the iterative design process driven by thermal simulations rendered on COMSOL Multiphysics. This process led to the conception of a 2-thruster and 4-thruster cluster frame. A lumped-parameter model of the electric propulsion system was also created to model its complex thermal behaviour. In addition, the 2-thruster frame was studied mechanically with analytical calculations and simulations of simple load cases on SolidWorks. Lastly, a prototype based on the 2-thruster frame model was assembled. The prototype was used to conduct temperature measurements while hosting two operating thrusters inside a vacuum chamber. The temperature distribution in the cluster was measured, and compared to simulation results. Thermal simulations of the 2-thruster and 4-thruster frame showed promising results, while mechanical simulations of the 2-thruster version met all requirements. Moreover, experimental results largely agreed with thermal simulations of the prototype. Finally, the lumped-element model proved instrumental in calibrating the models, with its high flexibility and quick computation time. / Generellt erbjuder elektrisk framdrivning hög verkningsgrad men relativt låg dragkraft. För att avhjälpa detta kan flera jonmotorer sättas samman i en klusterkonfiguration och drivs parallellt. Detta kräver en noggrann utformning av en ram för att rymma de enskilda framdrivningssystemen. Denna ram måste vara modulär för att kunna användas i olika klusterstorlekar och verifiera termiska och mekaniska krav för att säkerställa den nominella driften av motorerna. Föreliggande rapport syftar till att visa designprocessen för en sådan ram, från preliminär modellering till experimentell studie av en prototyp. Detta dokument innehåller en översikt över den iterativa designprocessen, driven av termiska simuleringar gjorda med COMSOL Multiphysics, som ledde till uppfattningen av en 2 motorer och 4 motorer ram. En klumpelementmodell av jonmotorn skapades också för att modellera dess komplexa termiska beteende. Dessutom var den 2 motorer ram studeras mekaniskt med analytiska beräkningar och simuleringar av enkla laddafall med SolidWorks. Slutligen monterades en prototyp baserad på den 2 motorer rammodellen. Prototypen användes för att göra temperaturmätningar medan den är värd för 2 jonmotorer i en vakuumkammare. Temperaturfördelningen i klustret mättes och jämfördes med simuleringsresultat. Termiska simuleringar av den 2 motorer och 4 motorer ramen visade lovande resultat, medan mekaniska simuleringar av den 2 motorer versionen klarade alla krav. Dessutom överensstämde experimentella resultat till stor del med termiska simuleringar av prototypen. Slutligen var klumpelementmodellen mycket användbar för att kalibrera de andra modellerna med sin höga flexibilitet och snabba beräkningstid.

Implanter une machine électrique dans un véhicule hybride pose avant tout des problèmes d’encombrement. Sous-dimensionner la machine semble légitime compte tenu de l’usage qui en est fait sur véhicule. Par contre, cela suppose que les aspects thermiques soient pris en compte non seulement lors de l’utilisation, mais aussi lors de la conception de la machine. Le phénomène majeur limitant la densité de puissance massique des machines électriques est l’échauffement interne des bobines et des aimants. La modélisation thermique de la machine est complexe compte tenu de la diversité des sources de chaleur et de la coexistence de différents modes de transferts thermiques : conduction dans la matière, convection avec l’eau de refroidissement, conduction, convection et rayonnement dans l’entrefer. En termes de géométrie, si une première approche peut être réalisée en ne considérant que des flux de chaleur radiaux, la composante axiale doit nécessairement être prise en compte dès lors qu’on veut tenir compte aussi des extrémités de machine, et notamment de la chaleur produite par les roulements et les têtes de bobines. Ainsi pour pouvoir analyser pertinemment les transferts thermiques dans la machine, des méthodes numériques de type CFD ont été utilisées pour caractériser le transfert thermique par convection. La caractérisation des échanges thermiques par convection naturelle et forcée a été réalisée à l’intérieur d’une machine synchrone à aimants permanents internes (MAPI). Des relations empiriques ont été proposées afin de prendre en compte le couplage entre la thermique et l’aérodynamique dans les cavités des machines électriques totalement fermées. Afin de valider la pertinence des modèles numériques dans le cadre de ce travail, des mesures thermiques à l’aide des moyens d’essais de l’UTC ont été réalisées. Les résultats de cette étude sont utilisés pour construire des circuits thermiques équivalents qui prennent en compte les phénomènes thermiques complexes dans les machines électriques fermées utilisées dans les véhicules hybrides et électriques. Ces recommandations de conception permettront l’optimisation de l’effort investi pour le refroidissement de la machine électrique dans ses différentes phases de fonctionnement. / The integration of an electrical machine into modern hybrid vehicles is associated with new technical constraints such as the integrability into small volume without losing certainly in performance. Therefore, the development of compacter electrical machines is a well-founded argument for car manufacturers as well as electrical machine designers. On the other hand, this finding assumes that the thermal aspects are undertaken not only during the operation of the electrical machine, but also during the design process. The internal heat generated in different areas impacts strongly the power density and the magnet health which deeply reduce the electrical machine reliability. Heat transfer modeling inside electrical machines is a tricky task because of the strong coupling between the different physics governing their operations. Indeed, the generated losses spread inside the electrical machine through three heat transfer modes which are: conduction (heat diffusion), convection(heat transport) and radiation (heat scattering). In terms of geometry, if a first approach can be carried out by considering only radial heat fluxes, the axially-transferred heat must be undertaken when it is also necessary to consider end caps effects, and particularly the heat released by the bearings. In order to carry out relevantly the thermal analysis of a permanent magnets synchronous machine, CFD based methods are used to characterize the convective heat transfer inside this machine over a large operating range. Both natural and forced convection are analyzed and the corresponding heat transfer coefficients are numerically-estimated. Empirical equations are proposed in order to take into account the coupling between thermal and fluid dynamics inside the cavities of the studied totally-enclosed machine. These correlations are integrated then into a detailed and reduced thermal network. Experimental tests are carried out using a test bench in order to measure temperature distribution in different areas of the electrical machine. Afterward, a comparison between estimated and measured temperatures shows that the results of the numerically-enhanced thermal network are in a good agreement with measurements. Thus, the proposed recommendations based on CFD modeling allow the convective heat transfer to be characterize quickly and precisely. These correlations are useful for upcoming studies dealing with convection inside automotive electrical machines as well as high speed electrical machines.

<div>This body of work examines analytical and numerical models to simulate the response of structures in shock absorption applications. Specifically, the work examines the prediction of cushion curves of polymer foams, and a topological examination of a $\chi$ shape unit cell found in architected mechanical elastomeric metamaterials. The $\chi$ unit cell exhibits the same effective stress-strain relationship as a closed cell polymer foam. Polymer foams are commonly used in the protective packaging of fragile products. Cushion curves are used within the packaging industry to characterize a foam's impact performance. These curves are two-dimensional representations of the deceleration of an impacting mass versus static stress. The main drawback with cushion curves is that they are currently generated from an exhaustive set of experimental test data. This work examines modeling the shock response using a continuous rod approximation with a given impact velocity in order to generate cushion curves without the need of extensive testing. In examining the $\chi$ unit cell, this work focuses on the effects of topological changes on constitutive behavior and shock absorbing performance. Particular emphasis is placed on developing models to predict the onset of regions of quasi-zero-modulus (QZM), the length of the QZM region and the cushion curve produced by impacting the unit cell. The unit cell's topology is reduced to examining a characteristic angle, defining the internal geometry with the cell, and examining the effects of changing this angle.</div><div>However, the characteristic angle cannot be increased without tradeoffs; the cell's effective constitutive behavior evolves from long regions to shortened regions of quasi-zero modulus. Finally, this work shows that the basic $\chi$ unit cell can be tessellated to produce a nearly equivalent force deflection relationship in two directions. The analysis and results in this work can be viewed as new framework in analyzing programmable elastomeric metamaterials that exhibit this type of nonlinear behavior for shock absorption.</div>

Les machines électriques jouent un rôle très important dans la conversion d'énergie dans plusieurs applications et domaines. Les contraintes thermiques jouent ainsi un rôle indispensable dans la conception des machines électriques de plus en plus petites et performantes. En effet, la performance des machines électriques est limitée par les températures maximales admissibles dans certaines zones critiques telles que le bobinage, les aimants permanents et les roulements. Deux approches principales peuvent être utilisées pour étudier le comportement thermique de la machine: la méthode nodale ou le circuit à constantes localisées ou les modèles numériques. Dans notre étude, nous proposons d'appliquer la méthode nodale sur une machine électrique intégrée dans un environnement mécatronique complexe. Le modèle thermique développé de la machine est ainsi présenté avec ses différents éléments. En effet, un modèle précis dépend fortement de plusieurs paramètres thermiques tels que les coefficients d'échange convectif, les conductances de contact, les conductivités équivalentes du bobinage, et autres paramètres. En conséquence, des techniques d'analyse de sensibilité sont ensuite appliquées sur le modèle thermique pour identifier les paramètres d'influence significative sur les températures de la machine ainsi que pour la réduction de ce modèle. Ensuite, nous appliquons deux méthodologies d'identification des paramètres thermiques incertains sont développées et appliquées afin de recaler le modèle thermique de la machine. Cette étape permet la validation de ce modèle par rapport à des mesures thermiques sur une machine synchrone à aimants permanents internes installée sur un banc de caractérisation de machine électriques. Finalement, nous intégrons le modèle recalé dans une approche système mécatronique comportant les lois de commande de la machine ainsi que son convertisseur. Ceci permettra ainsi d'étudier l'influence de la température d'une machine électrique sur le système mécatronique complet. / Electric machines play an important role in power conversion in several applications and fields. With the increasing demand for designing lighter and more efficient machines and optimizing the existing structures, thermal analysis becomes a necessary; in fact, the performance of electric machines islimited by the allowable temperatures in many critical components like windings, permanent magnetsand bearings. Two main approaches can be employed in order to study the machine thermal behavior : the lumped parameter thermal network (LPTN) or numerical models. Considering low-computationtime-consuming and the possibility to be integrated in a mechatronics system design, the LPTN method is considered in our study. The latter is mainly applied on electric machine integrated in a complex mechatronics environment. The thermal network is presented along with the definition of the principal elements constituting this network. In fact, an accurate and reliable network strongly depends on many critical parameters like heat transfer coefficients, interface gaps, impregnation goodness, among others. For this reason, different sensitivity analysis techniques are carried out in order to, first, identify the significance of uncertainties in the evaluation of these parameters on machine temperatures and second, to reduce the thermal network. Next, we propose two optimization algorithm-based identification methodologies in order to calibrate results of the thermal network with measured temperatures obtained from a test-bench of a permanent magnet based integrated starter-generator machine. The calibrated model is then integrated in a mechatronics system consisting of an electric model of the electric machine, along with its control strategy and the power converter. This final study allows us to evaluate the impact of the machine temperature rise on the mechatronic system.