High volume production of Electric Machines : Results from WP3 in FFI project P43352-1

(1)

High volume production of Electric Machines

Results from WP3 in FFI project P43352-1 Project Report 26267, 2019-10-10

Mats Werke, Alexander Mann & Tomas Luksepp - RISE IVF AB Christian Dubar & Oskar Wallmark – KTH

Johan Hellsing & Dan Hagstedt – CEVT Emil Gavling & Ali Rabei - AAM

(2)

(3)

RISE IVF AB P O Box 104 SE-431 22 Mölndal Telephone +46 (0)31-706 60 00 Fax +46 (0)31-27 61 30 www.swerea.se Project Report 26267 © RISE IVF AB

RISE – The Swedish Research Institute

RISE is the Swedish Research Institute and

innovation partner. In international collaboration

with industry, academia and the public sector, we

ensure the competitiveness of the business

community and contribute to a sustainable society.

Our 2,700 employees support and promote all

manner of innovative processes. RISE is an

independent, state-owned research institute that

offers unique expertise and about 100 testbeds and

demonstration facilities, instrumental in

future-proofing technologies, products and services.

www.ri.se

As of 1 October 2018, RISE Research Institutes of

Sweden AB is, indirectly, the only shareholder in

RISE IVF AB (former Swerea IVF AB).

(4)

1. Summary

This report has been made with the intention to study vehicle traction electric machine design and electric machine high-volume production simultaneously. The aim was to better understand the trade-offs between optimal machine design versus production costs.

Four permanent magnet electric machine designs, all different from a production point of view, has been designed. Each machine design has been optimized as far as the project budget allowed to reach the shortest design which would fulfill all technical requirements.

For each of the four electric machine designs, a highly automated production line has been set up, capable of producing one million electric machines per year. All the costs for incoming material and for running the production line has been estimated to the best information available.

The results show that material cost is clearly dominating over production cost for all four machine designs with at least a factor of 20. This leads to the very simple conclusion that vehicle traction electric machines can be selected based on material cost only.

2. Acknowledgements

The project participants want to send a warm thank to AAM in Trollhättan for hosting this sub-project within their main project MEDS, and to the Swedish Energy Agency for the financial support to be able to perform this study. We are also most grateful to the companies we were allowed to visit, and for the fruitful discussions around electric machine production we could enjoy during our visits:

ABB LV Motors, Västerås BEVI, Blomstermåla

Cogent Power, Surahammar Dahréntråd, Nossebro

Xylem Water Solutions, Emmaboda

3. Introduction

The solution to the major challenge for the transportation sector to reduce its impact on the global climate is today clearly pointing towards electric traction, much due to its superior efficiency and excellent controllability. Regardless if the electric energy arrives from the electric grid, and stored in on-board batteries, or generated on-board, by a fuel cell system or by a combustion engine and generator system, electric machines for vehicle traction will be the main propulsion system and will thus be produced in considerably larger volumes than today.

(6)

The purpose of this study is to give us a better understanding of the implications on production methods as well as electric machine design, in the context of high-volume production of electric vehicle traction systems. High high-volume is here defined as a production rate of 1 million identical electric machines yearly. This is high, but not unreasonable, considering several vehicle OEMs are producing 5-10 million vehicles per year.To keep this study fairly small, the plan was never to build a generic production cost model for all possible electric machine variants, but instead to design a fictive high-volume production line for a limited number of popular electric machine variants and see which conclusions can be drawn from this more limited scope. Also, as the vast majority (93% according to [1]) of electric machines for vehicle traction are based on high-energy permanent magnets, this study will actually focus on different variants of the Permanent Magnet Synchronous Machine, PMSM, only.

At the initiative of CEVT and AAM, two companies within gearbox design and electric machine integration, a team was gathered with participants from CEVT, AAM, KTH, and RISE. This study has been run as a sub project to the FFI project MEDS (Cost efficient, scalable and modular electric drive systems for next generation propulsion and lateral vehicle control).

CEVT, China Euro Vehicle Technology, Powertrain Engineering department, has contributed with chapter 4 “Method” and has been represented by Johan Hellsing (johan.hellsing@cevt.se) and Dr Dan Hagstedt (dan.hagstedt@cevt.se).

AAM, American Axle, Trollhättan office, has been represented by Emil Gavling (emil.gavling@aam.com) and Dr Ali Rabei.

KTH, Royal Institute of Technology, Division of Electric Power and Energy Systems, has contributed with chapter 5 “Electric machine design” and has been represented by Dr Christian DuBar and Associate Professor Oskar Wallmark (owa@kth.se).

RISE IVF AB, Research Institutes of Sweden, Department of Manufacturing Processes, has contributed with chapter 6 “Production costs” and has been represented by Dr Mats Werke (mats.werke@ri.se), Alexander Mann (alexander.mann@ri.se) and Tomas Luksepp (tomas.luksepp@ri.se).

4. Method

The plan was to investigate a limited number of electric machine designs, which shall be clearly different from each other in terms of production methods, and they shall still fulfill the same technical performance intended for vehicle traction. For each machine design, a production line for 1 million units per year shall be designed and for each machine design a total production cost shall be estimated. The production process of the PMSM stator contains clearly more steps than the production of the PMSM rotor. Also, in real vehicle applications, we have not found so much variations in rotor design affecting production. Thus, in order to save time, the investigated machine designs have been chosen based on varying stator design only. Initially in the project, it was decided to investigate the four stator concepts shown in Figure 1:

(7)

Concept 1. The conventional distributed winding with a multitude of individual thin round strands, inserted in a stator made out of steel sheets identical to the machine’s final shape. This is the most commonly used stator type in the kW-range of any kind of electric machine – PM or induction.

Concept 2. A “hair pin” distributed winding with a few square-shaped strands per slot, inserted in a stator made out of steel sheets identical to the machine’s final shape. The axially inserted square-shaped strands allow a very high slot fill factor, but every “hair pin” has to be welded to the next one.

Concept 3. A concentrated winding based on thin round strands, wound around each tooth of a straight stator, being folded it to its final round shape after winding. This method allows a better slot fill factor than concept #1.

Concept 4. A concentrated winding wound around a single tooth attached to part of the stator yoke. Winding impregnation is made before assembly to its final shape. All stator teeth are identical, which means even higher production volumes of each separate teeth. Should be well suited for highly automated production.

Figure 1: Conventional winding (upper left), Hairpin winding (upper right), Concentrated winding (lower left), Single tooth winding (lower right)

(8)

All of the above concepts shall fulfill the technical specification given in chapter 5, Table 1. The design process has been; optimizing stator as well as rotor for stator concept 1 first, and from there carry over the rotor design to the other three stator concepts and optimize each of the stators to this fixed rotor design. This was done in order to save time and is believed to lead to quite marginal errors on the final cost estimation. Machine length was of course allowed to adjust but current density has been held constant between the four concepts.

In case of rotor production, the same process has been assumed for all of the above four concepts: Slipping all rotor sheets onto a rotor shaft with a stable end plate on each side, fixing the rotor pack with a locking device, and finally inserting the pre-magnetized interior magnets in rotor slots and fixing with a glue.

To clarify, designs based on Induction machines (IM), Synchronous machines with separately excited field winding (SM) and Synchronous reluctance machines (SynRel) would all be perfectly feasible for vehicle traction but has in this study been omitted to narrow down the scope. Main reason is their low penetration rate in actual vehicle applications.

4.1 Cost model

The basic idea behind this report is to better understand the balance between electric machine design and electric machine production. We have chosen to do this by looking on the total cost situation of a fictive production facility capable of producing 1 million electric machines (stator + rotor) per year.

The balance between engineering design and production costs is found via the total production cost built up of material cost, mainly from incoming material and material scrap, and of production cost, mainly from investments in equipment, its maintenance and the production staff. This simple cost model is written as Eq. (1).

CEM = Cmtrl + Cman + Ceq + Cto – Crest (1)

where

Cmtrl = Cost of incoming material to factory

Cman = Cost of operators in manufacturing process

Ceq = Cost of manufacturing equipment

Cto = Cost of manufacturing maintenance and tools

Crest = Rest value of scrapped material

This is of course a simplification but is intended to cover the main costs of electric machine production. Costs not covered by Eq. (1) are for example cost for the production facility, administration, energy costs and costs for scrapped units (units not passing final test has a lower value than Crest). The resulting production cost

(9)

captured the cost difference between the investigated electric machine variants in a fairly correct manner.

4.2 Cost of inverter

As the four different machine variants has different peak phase currents to be able to produce the requested peak torque, the capability of the inverter will have to vary between the different machine variants. This will of course influence the cost of the inverter and thus the total cost of the electric traction system. The inverter cost is not been covered by our cost model, but the reader can, with the help of Table 6, estimate the differences in inverter cost between the different machine concepts by assuming a proposed variable cost of 100 SEK per each 100 A of peak phase current.

(10)

5. Electric machine design

This section describes the final electric machine designs. Only a few aspects have been treated in detail, but more technicalities can be found in the two conference papers [3,4] that have been published during the project. The two machine concepts with concentrated windings (concept 3 & concept 4) are, from an electromagnetic perspective, considered as identical machine designs.

5.1 Design requirements

The target is to determine the minimum required stack length in order to fulfill the design requirements, presented in Table 1, for each of the machine concepts. As the torque and speed demands are both specified for 30 minutes and 10 seconds duration of operation, thermal boundaries are needed. The thermal initial condition is somewhat simplified as all the parts in the machine are considered to be 100°C. The upper thermal limit is a maximum winding hotspot temperature of 180°C. Table 1: Design requirements.

Parameter Value DC-link voltage 320 VDC Speed 0-14000 rpm Torque (30 min) 80 Nm Torque (10 sec) 220 Nm Power (30 min) 40 kW (5000-12000 rpm) Power (10 sec) 110 kW (5000-10000 rpm) Torque ripple (p-p) <9% (±10 Nm @ 220Nm)

Stator outer diameter 180 mm

Initial temperature 100°C

Winding hotspot temperature Max 180°C

Liquid cooling (inlet) 70°C @ 6 l/min

5.2 Concept 1 – Conventional winding

The interior permanent magnet synchronous machine (IPMSM) design with the insert winding is illustrated in Figure 2, where individual strands are modeled in the two middle slots. The strands are placed randomly, and the eddy current losses are calculated in each strand separately (i.e. the surface integral of the current density in a single strand correspond to its portion of the phase current).

(11)

Figure 2: Geometry of concept 1 – insert winding, randomly distributed strands are modeled individually in the two middle slots.

The key design parameters of the Concept 1 machine design, with insert winding, are presented in Table 2. Exactly the same rotor and the same materials are utilized for the other concepts as well.

Table 2: Key design parameters of concept 1 together with the materials used.

Parameter Value

Number of strands per slot 56

Strand diameter 0.81 mm

Copper fill factor 0.4

Slot openings 1.88 mm

Pole pair number 4

Number of slots 48

Outer stator diameter 180 mm

Rotor diameter 120 mm

Air-gap length 0.7 mm

PM thickness 2.3 mm

PM width 15.9 mm

Shaft hole diameter 40 mm

Part Material

Stator & rotor M270-35A

Permanent magnets NdFe36

5.3 Concept 2 – Hairpin winding

The final geometry of the hairpin winding has been created in three steps, which is shown in Figure 3. In the first step, the keyhole shaped slot used for the insert winding is converted into a rectangular shaped slot that better suits the rectangular wires. The height of the shoe (a), the thinnest part of a tooth (b) and the yoke (c)

(12)

are all kept the same. Four strands are located in the resulting slots, considering a copper fill factor of 0.7. However, in [3] it was shown that the AC-copper losses at the maximum speed of 14000 rpm are almost ten times higher than the corresponding DC-copper losses. One way to reduce the losses at high speeds or frequencies, shown in [3], is to introduce an empty space closest to the slot opening (d). In the third step, the four strands and the yoke are moved back towards the slot opening the same distance (d). As the distance between two strands (0.18 mm) and the distance between the strands and the stator iron (0.38 mm) were both kept, the resulting copper fill factor becomes 0.63. Another consequence is that the outer stator diameter is reduced from 180 mm to 161 mm. It would of course also be possible to increase the number of strands in order to fill up the original slot to the expense of production complexity and new input requirements from suppliers of manufacturing tools. The latter was not possible within this project due to the restricted time.

Figure 3: Illustration of how the slot for the insert winding is modified to a slot suitable for the hairpin winding.

The final hairpin design is illustrated in Figure 4, where individual strands are modeled in the two middle slots in the same manner as for concept 1 with the insert winding.

Figure 4: Geometry of concept 2 – hairpin winding, rectangular strands are modeled individually in the two middle slots.

The key design parameters of concept 2 with the hairpin winding are collected in Table 3. Observe that the outer stator diameter of concept 2 is reduced from 180 mm to 161 mm.

(13)

Table 3: Key design parameters of concept 2.

Number of strands per slot 4

Strand width 1.7 mm (radial)

Strand height 2.3 mm (circumferential)

Number of slots 48

Outer stator diameter 161 mm

5.4 Concept 3 and 4 – Concentrated winding

The geometry of the 8 pole 12 slot design with a concentrated 2-layer winding, that represents both concept 3 and concept 4, is illustrated in Figure 5. The eddy current losses are modeled in individual strands in the middle slot, two layers of 24 strands belonging to two different phases (in total 48 strands in a slot). Exactly the same rotor design, air-gap length and also the slot opening as in the previous designs are used.

Figure 5: Geometry of concept 3 and 4 – concentrated winding, individual strands are modeled in the two middle slots.

(14)

Table 4: Key design parameters of concept 3 and 4.

Number of strands per slot 48 (24 per layer)

Strand diameter 2 mm

Number of slots 12

5.5 Thermal modeling

The thermal properties of the materials that are used in the transient thermal modeling of all the winding concepts are shown inTable 5. The thermal modeling is described more in detail in [4]. The heat conductivity of the permanent magnets (NdFe36) is not considered as the whole rotor is modeled as a single domain, which constitutes of electrical steel lamination, permanent magnets and air pockets. Table 5: Material data used for the transient thermal models.

Material Density (kg/m3)

Heat conductivity (W/(m·K))

Specific heat capacity (J/(kg·K))

M270-35A 7650 28 471

Copper 8933 400 391

NdFe36 7500 - 502

Insulation - 0.2 -

An example of the transient thermal responses of the winding hotspot temperature, for concept 2 in the operating point Jd= -12 ARMS/mm2 and Jq= 24 ARMS/mm2, at

different speeds are shown in Figure 6. It can be seen that the winding hotspot temperature always starts at 100°C according to the initial condition, then it increases or decreases depending on the operating speed. The horizontal, black and dashed line indicates the maximum winding hotspot temperature of 180°C. From Figure 6, it can be concluded that the machine can be operated for 30 minutes in the operating point, Jd= -12 ARMS/mm2 and Jq= 24 ARMS/mm2, for a speed of 10000 rpm

but not for a speed of 12000 rpm as the latter exceeds the maximum winding hotspot temperature at 30 minutes duration.

(15)

Figure 6: Transient winding hotspot temperature for concept 2 in the operating point Jd= -12 ARMS/mm2 and Jq= 24 ARMS/mm2, at different speeds.

When several operating points are evaluated for different speeds, as shown in Figure 6 previously, it is possible to represent the thermal limits in the JdJq-plane.

In Figure 7, the thermal limits of the maximum winding hotspot temperature for different speeds and durations of 30 minutes are shown. The operating point, Jd=

-12 ARMS/mm2 and Jq= 24 ARMS/mm2, is marked out with a black cross and is located

between the 10000 rpm and 12000 rpm limits in a more general context than in Figure 7, where it was presented in the time-domain. The same principle can be used in order to find the thermal limits for shorter or longer time durations and/or in other parts of the machine, in the permanent magnets for instance.

(16)

Figure 7: Thermal limits of transient operation during 30 min duration, considering a winding hotspot temperature of 180°C for concept 2.

5.6 Required stack lengths

The required stack lengths and the torque speed profiles are determined in two steps for all the designs:

1. The required stack lengths are determined for the 30 min and the 10 sec duration of operation up to 5000 rpm, purely based on the thermal limit (current limit) without taking the voltage limit into account. However, the current limit is determined by the speed as presented in the previous section. 2. The number of turns in the windings is increased as high as possible at the same time as the machines are ensured to be able to operate up to the design requirements, while the voltage limit is taken into account. This in order to reduce the required phase current as the DC-link voltage is fixed.

3. The shortest current vectors when considering 1. and 2. for a given operating point in terms of torque and speed are finally chosen.

A general observation that has been made is that the required stack lengths of all the machine/winding concepts are determined by the 220 Nm and 10 sec requirements.

The torque speed profile of concept 1 is shown in Figure 8, where the required stack lengths of 150 mm for 220 Nm/10 sec and 72 mm for 80 Nm/30 min are marked out. The maximum number of turns in this concept is limited by the operating point at 220Nm/10 sec at 5000 rpm which is pointed out by the green dot.

(17)

Figure 8: Required stack lengths and torque-speed profiles for concept 1.

The required stack lengths of 163 mm for 220 Nm/10 sec and 90 mm for 80 Nm/30 min for concept 2 are presented in Figure 9. The maximum number of turns in concept 2 is also limited by the operating point at 220Nm/10 sec at 5000 rpm which is shown by the green dot.

Figure 9: Required stack lengths and torque-speed profiles for concept 2.

In Figure 10, it can be seen that stack lengths of 159 mm and 79 mm are required for concept 3 and 4 when considering the 220 Nm/10 sec and the 80 Nm/30 min requirements respectively. The green dot indicates that the maximum number of turns in these concepts is determined by the 110 kW/10 sec operating point at 10000 rpm.

(18)

Figure 10: Required stack lengths and torque-speed profiles for concept 3 and 4. 5.7 Torque ripple

The peak to peak torque ripple in Figure 11 is calculated for the 220 Nm/10 sec requirement up to 5000 rpm, thereafter is the 110 kW/10 sec operating points followed up to 14000 rpm (although only up to 10000 rpm is specified in the design requirements). It can be noticed that both concept 1 and concept 2 fulfills the design requirement of 20 Nm (p-p) over the whole region. For concept 3 & 4, the torque ripple is about 23.5 Nm (p-p) up to 4000 rpm and it is below 20 Nm (p-p) from 4936 rpm up to 8185 rpm. After 8185 rpm, the torque ripple increases up to its maximum of 57 Nm (p-p) at about 13500 rpm.

Figure 11: Torque ripple when following the 10 sec requirement 0-14000 rpm. Step skewing of the rotor, in order to reduce the torque ripple, in concept 3 & 4 was investigated. Nevertheless, it was found that the rotor had to be skewed forwards

(19)

(magnet flux in the direction of the d-axis shifted towards the q-axis) in order to reduce the torque ripple to adequate levels. However, this is in principle equivalent to a reduction of the current angle without skewing. That means that the voltage limit is interfered, and the maximum number of turns is no longer valid. It should be pointed out that the study of a step skewed rotor was examined for operating points above 8185 rpm, which means that the machine is operated under field weakening and that the voltage limit is directly interfered. Finally, an alternative method was used to eliminate the torque ripple in concept 3 and concept 4:

1. Required stack length based on the thermal limits as before but with a torque ripple below 20 Nm (p-p).

2. As high number of turns as possible in order to fulfill the design requirements both in terms of torque, speed and a torque ripple below 20 Nm (p-p).

3. The shortest current vectors when considering (1) and (2) for a given operating point in terms of torque and speed are finally chosen.

When fulfilling the torque ripple requirement of 20 Nm (p-p) for concept 3 and concept 4, a stack length of 178 mm is required as pointed out in Figure 12. The maximum number of turns is now determined by the operating point 220Nm/10 sec at 5000 rpm, in the same manner as for concept 1 and concept 2.

Figure 12: Required stack lengths and torque-speed profiles for concept 3 and 4 when the torque ripple requirement is fulfilled.

The final torque ripple levels are shown in Figure 13, where the torque ripple of concept 3 and 4 have been suppressed below 20 Nm (p-p). It can be seen that the suppression of the torque ripple 0-5000 rpm is determining the stack length. Recall that the step skewing was not investigated in the 0-5000 rpm region, it is therefore still possible that step skewing could reduce the stack length slightly due to a torque ripple reduction in this region.

(20)

Figure 13: Final torque ripple when following the 10 sec requirement 0-14000 rpm. 5.8 Design summary

A design summary for all the different concepts is presented in Table 6, together with some figures for the Toyota Prius 2004 drive system. In [2], it is concluded that the inverter of the Prius 2004 drive system must be rated at least 169 kVA while the maximum power is limited to 50 kW. The power limit, in this case, comes from the available power from the DC-link (battery and generator together). The required size of the converter and the peak power are both interesting aspects. A highly utilized electric machine, made as small as possible to fulfill the 10 sec requirements, will potentially lead to a poorly utilized converter. On the other hand, it is theoretically possible to achieve more than 110kW in 10 sec for concept 1 (160kW) and concept 2 (145kW) in their respectively field weakening regions (see Figure 8 and Figure 9 again). However, it is likely that the power from the DC-link will be the limiting factor as in the Prius 2004 case.

(21)

Table 6: Design summary. #1 #2 #3 & #4 Tripple>20Nmp-p #3 & #4 Tripple<20Nmp-p Prius 2004

Stack length (30 min) 72 mm 90 mm 79 mm - -

Stack length (10 s) 150 mm 163 mm 159 mm 178 mm -

Stack volume 3.8 dm3 _{3.3 dm}3 _{4.0 dm}3 _{4.5 dm}3 _{4.8 dm}3

Peak power 160 kW 145 kW 131 kW 124 kW -

Peak power req. 110 kW 110 kW 110 kW 110 kW 50 kW1

Inverter rating 240 kVA 191 kVA 264 kVA 444 kVA 169 kVA

( kW : kVA) 1:2.2 1:1.7 1:2.4 1:4.0 1:3.38

Peak phase current 866 A 689 A 953 A 1602 A

The torque ripple issue for concept 3 and 4 was discussed several times during the project. On the one hand, we said that all technical requirements should be fulfilled – including the torque ripple requirement of course. On the other hand, we also realize that the torque ripple requirement might not be as well documented as the rest of the technical requirements. Still, the automotive industry knows very well how to treat powertrains with much larger torque ripples than this. The increased torque ripple over 8500 rpm we estimated should be less noticed by a driver as high motor speeds only happens at high vehicle speeds. Finally, it was decided to accept a “deviation” on the torque ripple and for the rest of this report we use only the 159 mm stator length for concepts 3 and 4 instead of the 178 mm stator length.

6. Production costs

6.1 General assumptions

The conventional stator concept, Concept 1, is well known in industry. One disadvantage is the coil winding process that requires several steps including pressing and lacing machines, besides the advanced winding and insertion of the coils. The hairpin concept, Concept 2, is a more recent development and gaining in popularity. However, the insertion and bending of the hairpins is complex. The concentrated winding concepts, Concepts 3 and 4, are clearly easier with regards to winding. However, the stator stack manufacturing will in turn pose greater challenges. There are advantages and disadvantages with the different concepts. The manufacturing process steps are illustrated in Table 7.

(22)

Table 7: Production processes for the investigated concepts

Process steps #1 #2 #3 #4

Stamping Whole stator Whole stator Straight stator Single tooth Core assembly Interlocking Interlocking Interlocking Backlacking Slot/tooth insulation Paper Paper Powder coating Powder coating Coil winding 1 coil per 2 slots Prefab hairpins Needle winding Spindle winding Coil insertion Fully automatic Insert & bending N/A N/A End winding Lacing, forming Welding, insulating N/A N/A Impregnation Trickling Trickling Trickling Trickling Final assembly N/A N/A Rolling Picking in place Phase connections Crimping Welding Welding Welding

The cost estimations are based on a production rate of 1 million machines per year. The production is to be placed in a low-wage country resulting in an operator cost of 35 SEK/h [6]. Three work shifts are assumed resulting in 6000 work hours per year. Depreciation time of equipment is set to five years. The parameters that were used to estimate the cost of each production stage are explained in the appendix.

6.2 Concept 1 and 2 – Stamping

The press line process steps are illustrated in Figure 14.

Figure 14: Press line process steps

Electric steel: The input to the press line is non-oriented electrical steel coil. Thick

coil is hot rolled (2 – 3 mm thick) and further processed using e.g. annealing in several steps, cold rolled to final thickness (e.g. 0,35 mm), shot blasted in order to remove oxides and finally a slitting process in order to obtain the desired width. A typical coil (e.g. M270-35A) consists of 3.6% Si,1.0% Al, 0.1% Mn and <0.003% C. Electrical steel is usually coated in order to increase electrical resistance between laminations to reduce eddy currents, provide resistance to corrosion or rust, and to act as a lubricant during die cutting. There are various coatings (organic and inorganic), and the coating used depends on the application

(23)

of the steel. Typical coatings are Suralac 7000 and Suralac 9000. Suralac 9000 may be heated to enable lamination stacks to bond together. This type of bonding delivers improved noise performance for large volume production of thin materials and is compatible with other traditional insulation coatings. When the coatings can adhesively bond the laminates together, they have been named as backlacking varnishes or just backlacking and is described in chapter 6.5.

The concepts use the non-oriented electric steel, M270-35. To calculate Acoil a

stamping layout was created using computer-aided design (CAD). A nested layout was chosen to reduce scrap. The laminates were placed so that the smallest distance between cutouts and between the cutouts and the edge equals 10 mm. The layout resulted in an electric steel coil width of 365 mm for concept 1 and 329 mm for concept 2. The amount of steel (Mmtrl Steel, Kg) that was needed per stator was

calculated using Eq. (2). The amount of scrap material (Mrest Steel, Kg) wascalculated

using Eq. (3). The cost of the material was estimated to 10 SEK/kg using [5]. The scrap material was assumed to be sold at 2 SEK/kg.

𝑀_{𝑚𝑡𝑟𝑙 𝑆𝑡𝑒𝑒𝑙} = 𝐴_{𝑐𝑜𝑖𝑙} ∗ 𝑙_{𝑠𝑡𝑎𝑐𝑘}∗ 𝐷_𝑒𝑠 (2) 𝑀_{𝑟𝑒𝑠𝑡 𝑆𝑡𝑒𝑒𝑙}= (𝐴_{𝑐𝑜𝑖𝑙}− 𝐴_{𝑠𝑡𝑎𝑡𝑜𝑟}− 𝐴_{𝑟𝑜𝑡𝑜𝑟}) ∗ 𝑙_{𝑠𝑡𝑎𝑐𝑘}∗ 𝐷_𝑒𝑠 (3)

Process: The press line starts with decoiling and feeding of the steel coil (t = 0.35

mm) into a stamping press. A mechanical press (3100 kN) with flywheel eccentric shaft was selected for concept 1 and 2. Progressive stamping, where the coil is stamped and moved forward in several steps, is assumed for concept 1 and 2, see Figure 15.

Figure 15: Progressive stamping

In order to improve material utilization both stator and rotor should be stamped in the same progressive tool. This is possible for concept 1 and 2 as they consist of circular laminates and the stator and rotor laminates have the same thickness. The stamping of nested stator laminates is recommended, see Figure 16.

(24)

Figure 16: Stamping arrangements for concept 1 and 2

After the stamping process the laminates are stacked to the desired thickness and assembled. The stack assembly can be performed using interlocking (dimples that are stamped out in the press), welding, bolts or backlacking technique. The waste material from the stamping is collected for recycling. In this project interlocking was assumed for concept 1 and 2.

Figure 17: Stator assembly using clips and bolts

The process parameters e.g. vstamping (250 spm), Nsheets (2 stator and rotor laminates

per hit) and U (90%) were given by a press manufacturer as well as the purchasing cost of a stamping line (23 MSEK) and tool costs (4.2 MSEK/year). The total number of laminates (Nlaminates) that needs to be produced per year were calculated

using Eq. (4). The yearly capacity (Ccap, laminates per year) of one stamping line was calculated using Eq. (5). The number of stamping lines required (three) were calculated by dividing the total number of sheets with the stamping line capacity. It was estimated that one operator was needed for each stamping line. It was decided that the stator and rotor carry half of the cost each for the stamping process.

𝑁_{𝑙𝑎𝑚𝑖𝑛𝑎𝑡𝑒𝑠} = 𝑙𝑠𝑡𝑎𝑐𝑘

𝑡_{𝑙𝑎𝑚𝑖𝑛𝑎𝑡𝑒}∗ 𝑁𝑢𝑛𝑖𝑡𝑠 (4)

(25)

6.3 Concept 3 – Stamping

As the stator laminates of concept 3 were stamped straight it was not suitable to stamp the stator and rotor laminates in the same process.

Electric steel: The same electric steel is used as in concept 1 and 2. Stator laminates

were positioned parallel in the stamping layout while the rotor laminates were placed in a nested layout. The smallest distance between cutouts and between cutouts and the edge was set to 10 mm for the both the stator and rotor layout. This resulted in an electric steel coil width of 77 mm for the stator laminates and 253 mm for the rotor laminates. The stamping layouts were used to calculate Acoil for

the stator and rotor laminates separately. The ingoing material was calculated using Eq. (2) for both the stator and the rotor. The amount of scrap material was calculated using Eq. (6) and Eq. (7).

𝑀𝑟𝑒𝑠𝑡 𝑆𝑡𝑎𝑡𝑜𝑟 = (𝐴𝑐𝑜𝑖𝑙− 𝐴𝑠𝑡𝑎𝑡𝑜𝑟) ∗ 𝑙𝑠𝑡𝑎𝑐𝑘∗ 𝐷𝑒𝑠 (6)

𝑀𝑟𝑒𝑠𝑡 𝑅𝑜𝑡𝑜𝑟 = (𝐴𝑐𝑜𝑖𝑙− 𝐴𝑟𝑜𝑡𝑜𝑟) ∗ 𝑙𝑠𝑡𝑎𝑐𝑘 ∗ 𝐷𝑒𝑠 (7)

Process: The same press was used as for concept 1 and 2 for both stator and rotor

laminates. The process parameters were set to 250 strokes per minute and 2 sheets per hit for the stamping of the stator laminates and 250 strokes per minute and 4 sheets per hit for the stamping of the rotor laminates. The number of required stamping lines were calculated as for concept 1 and 2 resulting in three lines to produce the stator laminates and two lines for the rotor laminates. It was estimated that one operator was needed for each stamping line. To bond the laminates to each other interlocking was used.

(26)

6.4 Concept 4 – Stamping

Because of the sectioned stator laminates (single tooth) in concept 4, the stator and rotor laminates were stamped in two separate processes.

Electric steel: The stator laminates are manufactured using an electrical steel with

a bonding coating on both sides. The rotor laminates use same electrical steel as in the other concepts. The sectioned stator laminates were placed in a row where every other section is rotated 180 degrees in the stamping layout. The smallest distance between cutouts and between cutouts and the edge was set to 5 mm for the stator laminates. This resulted in an electric steel coil width of 42 mm. The same rotor stamping layout was used as in concept 3. Acoil and the amount of scrap material

was calculated as in concept 3. The cost of the stator laminate material was estimated to 13 SEK/kg to include the more expensive backlacking coating.

Process: A linear type die making was assumed. The stator laminates were stamped

in a smaller press (1000 kN) than in the other concepts. The process parameters e.g. vstamping (450 spm), Nsheets (4 stator section laminates per hit) and U (90%) were

given by a press manufacturer as well as the purchasing cost of a stamping line (7 MSEK) and tool costs (3 MSEK/year). To stamp the rotor laminates the same press was used as in the other concepts. The process parameters were set as in concept 3. The number of stamping lines was calculated as before, resulting in ten lines to produce the stator laminates and two lines to produce the rotor laminates. To bond the stator section laminates together, backlacking was used. A yearly cost for backlacking was estimated to 8.5 MSEK/year, including equipment and operator costs. This cost is included in the stamping production cost.

6.5 Concept 4 – Backlacking

Backlacking can be used for stacking of stator sheets. The backlacking technology give advantages. The joining of the lamination stack will not disturb the magnetic field as interlocking can do. This can both save material, reduce weight and costs for the manufacturing. Bonded stators also give a noise reduction, better thermal conductivity between sheets and improved corrosion resistance.

Three European suppliers of the backlacking technology were easily found, see table 9.

- Stabolit 70 Thyssen Krupp

- Suralack 9000 Cogent/Tata (Surahammars bruk) - Voltatex 1175 WL (WH) Axalta coating systems

Despite differences in the technical datasheets the products seem to be very similar. The varnishes are based on waterborne epoxy resins. They are applied on one or two sides of the coil. After drying the typical thickness is approximately 5 µm. In the as-delivered condition the varnish is completely dry and non-adherent. Sheets can be stacked and coils to be tightly wound without any adhering or bonding at normal storage temperatures. Even so after pressure from high weights and

(27)

extended period of storage. The varnish lubricates at punching and prolongs the life of the stamping tools.

Table 8: Key parameters from three backlacking supplier’s official data sheets.

Suralac 9000 Stabolit 70

Voltatex 1175 WL & WH

Low & high viscosity

Base Epoxy Epoxy Epoxy (waterbased)

Colour Clear Clear

Surface Dry Dry and non-adherend Dry

Lubrication No oil at stamping No oil at stamping

Coating (µm) 4,5 5 (5-8) 0,5-7 Resistance ASTM A717 (Ω cm2₎ >200 5-500 Bonding pressure (N/cm2₎ 300 150-300

Contact Contact possible

At a raise of temperature, the resin in the varnish will melt and go through a viscosity minimum before the start of cure, then viscosity increases due to crosslinking reactions making the polymer chains longer. The crosslinking reaction rate for epoxy systems usually start to become significant at about 140 - 150°C. Both induction and furnaces can be used for heating the laminated stacks. The epoxy resin needs a certain time and temperature to fully cure. The range can be wide. As example, it can be 2 hours at 140°C or 2 minutes at 200°C. This is object temperature. All parts of the product must reach this temperature level. At the same time there is an upper time and temperature limit where the degradation of epoxy resins begins. It is typically around 2 hours at 200°C or 2 minutes at 230°C. When heating a product, there can be large variations in temperature and temperature gradients in the product during heating and cooling. The exact temperature profile naturally depends on the stack geometry and heating method and must be tested in each individual case. The coldest part of the object must at least have a minimum curing cycle while the hottest part must not exceed the maximum temperature limits. The curing process can be very fast for small parts that can be fast heated to a high peak temperature and then fast cooled. For larger parts, longer time and lower peak temperature must be used to allow uniform heating inside minimum and maximum limits.

The heating is faster with induction than in furnaces, however, the preferred way depends on how the object can be heated and the pressure put on, together with rate of production.

Pressure is applied on the stack to get the laminates in good contact. One sided coating may work good for small stacks with good tolerances. With two sided coatings it is less risk for defects and poor gap filling when the available amount of varnish is larger, and each metal side already have a full coverage.

A pressure of 150 to 300 N/cm2_{is recommended. Pressure must be applied from}

(28)

process. Spring back in the stack before the strength has been fully developed can lower the quality of the bonds. Under stated conditions the shrinkage per lamination and side is 2.5 to 3.5 µm.

Over the width the metal strips have some deviation in thickness, (~10µm) micrometer. To avoid buildup of tolerances and to limit the variations in gap thickness are each laminate can be rotated somewhat at assembly to even out this thickness variation.

Table 9: Curing data recommendation from three suppliers.

Temperature (°C) Suralac 9000 Stabolit 70 Voltatex 1175 WL

230 A few minutes 220-240 peak material temperature to reach 220 3-60 minutes 200 2 minutes 180 60 minutes 150 1-24 hours 140 2 hours Shrinkage (µm) per lamination and side

2,5-3,5

The bonded stacks can withstand temperatures up to 150°C during continuous operation. For a short period of time temperatures up to 200°C can be accepted without any damage.

The resin will have a glass transition temperature (Tg) around 100°C i.e. the material is still crosslinked but becomes softer and more flexible. The strength of the bonding is reduced by roughly 50% above this temperature. The strength will be fully recovered at temperatures below Tg.

The fully filled gaps in-between laminates will have a clear positive effect on the possibility of water ingression and the corrosion resistance.

In this type of applications, the strength is normally tested by peel testing methods. The result is then very dependent upon the thickness of the substrate tested. The test is most used for checking quality. The strength is in general far more than needed for the function of an electrical machine. By time there will be some degradation of the bonded joints. A fail-safe design is recommended.

One alternative bonding solution for the laminate stack is the Glulock technique

provided by Kienle + Spiess Group.

In this case the steel sheets have conventional isolation coatings and droplets of adhesive are applied in the stamping line. The droplets can be positioned in a free configuration. It is a two-component system most likely based on acrylate technology. The curing agent, hardener is sprayed on the opposite side of the panel. When two plates are stacked the base and hardener get in contact and the curing begins. Acrylate adhesives can be formulated to react very fast, down to about 1

(29)

min to full cure. There are little demands of a balance between the amount of base and hardener. Radical polymerization needs just initiation, so very small amount of hardener spray on the rear side of the plate is enough. The reaction rate can likely be adjusted to give time to fill and compress the stack in time before the open time of the adhesive has passed.

This bonding process gives no isolation coating. It is a spot bonding process that have the great advantage of not needing facilities for heat curing. Improved noise reduction should be the same as for backlacking, but the corrosion resistance and thermal conductivity should not reach the same level. Acrylates are known to have a good adhesion to most substrates and that even if some contamination is present. Normally, acrylates have a little lower heat resistance than epoxy resins, However, this Glulock HT (High Temperature) adhesive shall withstand 180°C, (class H). Depending on design and requirements, both Glulock or Backlacking could be the preferred choice for the most cost-effective solution. The stamping costs are summarized in Figure 19 – 20 and table 10 below.

(30)

Figure 19: Unit cost for the stator stamping process

Figure 20: Unit cost for the rotor stamping process Table 10: Total unit cost for the stamping process.

#1 #2 #3 #4

Stator 196 SEK 176 SEK 272 SEK 379 SEK Rotor 196 SEK 176 SEK 201 SEK 201 SEK

6.6 Concept 1 – Coil winding

Copper: Round copper wire is used in concept 1. Copper plates are melted, casted

and hot rolled down to 8 mm diam. The thick wire is heated up (> 300˚C) using e.g. a short circuit procedure, hot drawed down to 2.5, 1.5, 0.95 and 0.65 mm diam and then cold drawed to the final diameter (0.09–0.6 mm) in several steps. The wire is

-50 SEK 0 SEK 50 SEK 100 SEK 150 SEK 200 SEK 250 SEK 300 SEK 350 SEK 400 SEK 450 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

CEM - Stamping stator

Cto - cost of maint & Tools Cman - cost of operators Ceq - cost of equipment Crest- value of unused mtrl Cmtrl - material cost -50 SEK 0 SEK 50 SEK 100 SEK 150 SEK 200 SEK 250 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

CEM - Stamping rotor

Cto - cost of maint & Tools Cman - cost of operators Ceq - cost of equipment Crest- value of unused mtrl Cmtrl - material cost

(31)

then annealed in a furnace (350–400˚C) with protective atmosphere and varnished. During varnishing the wire passes through a varnishing resin and thereafter a varnishing steel matrix. The wire is passing through the resin and the steel matrix 10–20 times resulting in 10–20 varnishing layers on the wire. After varnishing, the coil is reheated in a furnace (400–750˚C) for removal of solvent and stress relief annealing. Finally, the wire is cooled for hardening and a thin layer of paraffin is applied for improving of the winding capacity.

Figure 21: Processed Copper wire

In order to calculate the amount of copper inside the stator slots Eq. (8)was used. To establish the amount of copper that was outside of the stator core (end windings), data from [7] was used. The report presents geometrical and mass properties of four different motors. The properties together with an assumed fill factor of 0.4 were used to calculate the amount of copper in the end windings in relation to the total amount of copper. The results are shown in table 11. The geometry of concept 1 was most similar to the LS 600h motor and therefore it was assumed to have the same proportion of copper in the end windings (55 %). The percentage together with the mass of copper inside the stator stack was used to calculate the total mass of copper. The cost of copper wire (100 SEK/kg) was estimated from [5]. No scrap material was considered.

𝑀𝑚𝑡𝑟𝑙 𝐶𝑜𝑝𝑝𝑒𝑟 = 𝐴𝑠𝑙𝑜𝑡∗ 𝑙𝑠𝑡𝑎𝑐𝑘∗ 𝑁𝑠𝑙𝑜𝑡𝑠∗ 𝐹 ∗ 𝐷𝑐𝑢 (8)

Table 11: Percentage of copper in end windings for different machines. Prius 2010 LS 600h Camry Prius 2004 Copper outside slots 72.3 % 54.8 % 70.9 % 70.8 %

Insulation: The insulation of the stator slots was considered as a part of the coil

winding process. The paper insertion into the slots, see Figure 22, is performed using a paper insertion machine.

(32)

Figure 22: Paper insertion

The amount of electrical insulation paper needed per stator (Ainsul, m2) was

calculated using Eq. (9). The 10 mm added to the stack length was an assumed stick out. The manufacturer of the insulation paper recommended a suitable thickness (0.25 mm) and a supplier of electric insulation material delivered a cost (135 SEK/m2).

𝐴𝑖𝑛𝑠𝑢𝑙 = 𝐶𝑠𝑙𝑜𝑡∗ (𝑙𝑠𝑡𝑎𝑐𝑘+ 10) ∗ 𝑁𝑠𝑙𝑜𝑡𝑠 (9)

Process: The concept 1 copper winding line is illustrated according to Figure 23.

(33)

The copper wire is wound into bundles where the bundles are then applied on the stator using a coil insertion tool, see Figure 24.

Figure 24: Coil insertion tool and coil insertion

In order to reduce the volume of coils outside the stator core the coils are pressed against the stator core in several steps using forming machines and fixated using a lacing machine, see Figure 25.

Figure 25: Preforming and lacing

The stator profile geometry and wiring diagram was delivered to a winding machine company. They could, using this information, deliver a purchasing cost (21.7 MSEK), operators per line (four) and capacity (55 stators per hour) for a complete coil winding line. The given capacity resulted in four coil winding lines.

6.7 Concept 2 – Coil winding

Copper: The amount of copper inside the stator stack was calculated as in concept

1. To establish the amount of copper in the end windings a 3D-model of the hairpin windings was created in CAD based on the geometry of the stator. The volume of the end windings in the 3D-model was then measured. The same cost for copper wire was used as in concept 1 (100 SEK/kg). No scrap material was considered.

(34)

Insulation: To insulate the stator slots insulation paper was used and the amount

needed was calculated the same way as in concept 1 except for an overlap of 2 mm. For concept 2 a thinner paper (0.18 mm) was recommended resulting in a lower price (86 SEK/m2).

Process: The hairpin winding concept is illustrated in Figure 26.

Figure 26: Hairpin winding process

A 3D-model and a wiring diagram was delivered to a winding machine company specialized in hairpin manufacturing. They could, using this information, deliver a purchasing cost (19.3 MSEK), operators per line (one) and capacity (35 stators per hour) for a hairpin winding line. The given capacity resulted in five hairpin winding lines.

6.8 Concept 3 and 4 – Coil winding

Copper: To calculate the amount of copper inside the stator stack the same method

was used as in previous concepts. As in concept 2 a model was created in CAD to determine the amount of copper in the end-windings. The same cost for copper wire was used as in previous concepts (100 SEK/kg). No scrap material was considered.

Insulation: Powder coating was used to insulate the stator. A manufacturer

recommended a suitable powder resin and the price was given by a reseller (613 SEK/dm3). In order to determine the volume of resin needed the area of all outer surfaces of a complete stator stack was measured in CAD and multiplied with a resin thickness of 0.25 mm.

Process: The winding process for concept 3 is a needle winding process, where the

iron core is fixed and the needle travels through the slots carrying the copper wire. The winding process for concept 4 is spindle winding. Here the stator tooth is mounted on spindle and where the copper wire is applied on the tooth by rotating the spindle. No cost estimation of the coil winding equipment was given by any of the consulted companies. The equipment cost for concept 3 and 4 only includes the powder coating lines. The coil winding costs are summarized in Figure 27 below.

(35)

Figure 27: Total unit cost for the coil winding process 6.9 Impregnation for all concepts using trickling

Trickling is a process where the pre heated components are held by special grabs and are set in rotation around their longitudinal axis. The resin is being trickled simultaneously on several parts of the wire winding of the rotating component at a present trickling temperature, see Figure 28. The windings are filled as much as possible with the resin. After the trickling process, the components are being cured, while still rotating, in a downstream heating process to obtain a constant and optimal impregnation result.

Figure 28: Trickling process

Resin: It was assumed that the resin would fill 100 % of the voids in the stator slots

and therefore have a filling grade of 1 – F of the total slot volume. No resin was considered outside of the slots. To calculate the amount of resin per stator (VResin,

dm3_{) Eq. (10) was used. A suitable resin was recommended by a manufacturer who}

also delivered a cost estimation (104 SEK/dm3).

𝑉_{𝑅𝑒𝑠𝑖𝑛} = 𝐴_{𝑠𝑙𝑜𝑡}∗ 𝑙_{𝑠𝑡𝑎𝑐𝑘}∗ 𝑁_{𝑠𝑙𝑜𝑡𝑠}∗ (1 − 𝐹) (10)

Equipment: Drawings of all concepts and a design summary was sent to a

manufacturer of impregnation equipment. Based on concept 2 they made the following estimations; 21.5 MSEK purchasing cost per trickling line, one operator per three lines, a capacity of 66 stators per hour per line and a machine availability of 85 %. The given capacity and machine availability results in 3 trickling

0 SEK 50 SEK 100 SEK 150 SEK 200 SEK 250 SEK 300 SEK 350 SEK 400 SEK 450 SEK 500 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

CEM - Coil winding stator

Cto - cost of maint & Tools Cman - cost of operators Ceq - cost of equipment Crest- value of unused mtrl Cmtrl - material cost, insulation Cmtrl - material cost, copper

(36)

impregnation lines. Although the estimates were based on concept 2, they could be applied to all concepts according to the manufacturer. The total impregnation unit cost is illustrated in Figure 29.

Figure 29: Total unit cost for the impregnation process 6.10 Magnets for all concepts

Magnets: The mass of the magnets that was needed per rotor (Mmtrl magnet, kg) was

calculated using Eq. (11). Since all concepts used the same rotor profile geometry the differences in cost derives from their stack lengths. The type of magnets used were NdFe36 with a temperature rating of UH (180° C) and a BHmax of 36. The

temperature rating and BHmax results in a concentration of 18 % neodymium (Nd)

and 7.5 % Dysprosium (Dy). The price for Nd (800 SEK/kg) and Dy (2500 SEK/kg) was found in [5]. Combining the concentration with the price of the materials resulted in a cost of 332 SEK/kg for the magnets. Adding the cost of iron, manufacturing, shipping etc. a price of 600 SEK/kg was assumed.

𝑀_{𝑚𝑡𝑟𝑙 𝑀𝑎𝑔𝑛𝑒𝑡} = 𝐴_{𝑚𝑎𝑔𝑛𝑒𝑡}∗ 𝑙_{𝑠𝑡𝑎𝑐𝑘}∗ 𝐷_𝑚 (11)

Equipment: The cost of the magnet assembly equipment was assumed to be low

and was therefore not investigated.

CEM - Impregnation stator

Cto - cost of maint & Tools Cman - cost of operators Ceq - cost of equipment Crest- value of unused mtrl Cmtrl - material cost

(37)

Figure 30: Material unit cost for the magnets

6.11 Total production costs

The total production costs are summarised below and concluded in chapter 7.

Figure 31: Total unit cost for each concept which includes all costs (material, equipment, operator and maintenance) of all production stages.

CEM - Magnets rotor

Cmtrl - material cost 0 SEK 200 SEK 400 SEK 600 SEK 800 SEK 1 000 SEK 1 200 SEK 1 400 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

CEM - total

CEM - Magnets Rotor CEM - Stamping Rotor CEM - Impregnation Stator CEM - Coil Winding Stator CEM - Stamping Stator

(38)

Tabell 12: Weight of each ingoing material for each concept. Note that some of the electrical steel will be scrap.

Material #1 (ref) #2 #3 #4

Electrical steel stator 19.87 kg 17.54 kg 26.50 kg 27.05 kg Electrical steel rotor 19.87 kg 17.54 kg 19.97 kg 19.97 kg Insulation stator 0.08 kg 0.04 kg 0.13 kg 0.13 kg

Copper stator 4.16 kg 1.62 kg 3.26 kg 3.26 kg

Resin stator 0.36 kg 0.08 kg 0.28 kg 0.28 kg

Magnets rotor 0.66 kg 0.72 kg 0.70 kg 0.70 kg

Figure 32: Total material cost per unit. Ingoing material (Cmtrl) with the return of

the scrap (Crest) subtracted. 0 SEK 200 SEK 400 SEK 600 SEK 800 SEK 1 000 SEK 1 200 SEK 1 400 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

Material

Magnets rotor Electrical steel rotor Resin stator Insulation stator Copper stator Electrical steel stator

(39)

Figure 33: Total operator cost (Cman) per unit.

Figure 34: Total equipment cost (Ceq) per unit (investment cost divided by

depreciation time). 0 SEK 1 SEK 2 SEK 3 SEK 4 SEK 5 SEK 6 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

Operator

Stamping rotor Impregnation stator Coil winding stator Stamping stator 0 SEK 10 SEK 20 SEK 30 SEK 40 SEK 50 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

Equipment

Stamping rotor Impregnation stator Coil winding stator Stamping stator

(40)

Figure 35: Total maintenance and tool cost (Cto) per unit. Only the stamping

processes are shown as Cto of the other production stages was not

specified by the machine suppliers.

7. Conclusion

The main conclusion to be drawn is the vast difference in material cost and manufacturing cost, respectively. This leads to the conclusion that the electric machine with the lowest material cost will also be the cheapest one to produce. Of course, some understanding of the production is necessary also for the machine designer as the production method can have some influence on material utilization, for example during stamping. But the manufacturing equipment has a close to negligible cost compared to material cost.

Therefore, the hair pin concept (#2) is shown here to fulfill the vehicle traction requirements at a 24% lower production cost than the reference (as well as a 3 kg lower weight)

The straight stator with concentrated windings (#3) has the lowest manufacturing cost, but due to higher material cost it does not come out as a cheaper product. The single-tooth concept (#4) looks very attractive, mainly in terms of material utilization during stamping and easily automated winding process. It still cannot be recommended, again due to material cost. The concepts with concentrated winding (#3 and #4) should of course not be ruled out in other applications as they both do look attractive in view of production technology. However, they are not recommended for the vehicle traction application.

In terms of sensitivity, no study has been made. But based on just the number of parallel production lines used, we can estimate that the results shown in this report could be valid for a production volume down to 300,000-400,000 electric machines per year. Below that, the manufacturing equipment will start to play a bigger role for the total production cost.

0 SEK 5 SEK 10 SEK 15 SEK 20 SEK 25 SEK 30 SEK 35 SEK 40 SEK 45 SEK #1 (ref) #2 #3 #4 C o st (SEK/u n it) Concept

Maintenance & tool

Stamping rotor Stamping stator

(41)

Looking more on the complete electric drive system by taking also the inverter cost into consideration should not change the above conclusion. As the hair pin concept has the lowest phase currents for peak operation, reasonably the inverter cost should also be the lowest.

For a vehicle level analysis, the implications on battery size should also have to be considered. This would require an analysis of the energy consumption of the different electric machine concepts. This has not been done in this study.

8. References

1. https://www.adamasintel.com/93-percent-evs-used-pm-motors-2018/ 2. Du-Bar, C. (2016). Design and analysis of a fault-tolerant fractional slot

PMSM for a vehicle application. Gothenburg, Sweden: PhD thesis,

Chalmers University of Technology.

3. Du-Bar, C., & Wallmark, O. (2018). Eddy current losses in a hairpin

winding for an automotive application. International Conference on

Electrical Machines (ICEM18). Alexandroupoli, Greece.

4. Du-Bar, C., Mann, A., Wallmark, O., & Werke, M. (2018). Comparison of

Performance and Manufacturing Aspects of an Insert Winding and a Hairpin Winding for an Automotive Machine Application. 8th

International Electric Drives Production Conferense (EDPC). Schweinfurt, Germany.

5. Fyhr, Pontus (2018) Electromobility Materials and Manufacturing

Economics, Lund, Sweden: PhD thesis, Lund University Faculty of

Engineering

6. Industrins arbetskraftskostnader internationellt, Teknikföretagen 2017 7. Burress, Campbell, Coomer, Ayers, Wereszczak, Cunningham, Marlino,

Seiber, and Lin Evaluation of the 2010 Toyota Prius Hybrid Synergy Drive

(42)

Appendix

Name Abbreviation Value/unit Explanation

Stack length lstack mm Total length of a complete stator

and rotor stack. Given by the electric machine design, see Table 6.

Laminate thickness tlaminate 0.35 mm The same value for all concepts,

both stator and rotor laminates.

Stator area Astator #1: 10296 mm2

#2: 7592 mm2

#3: 10458 mm2

#4: 10458 mm2

The area of the stator profile.

Rotor area Arotor 9383 mm2 The area of the rotor profile. The

area is equal for all concepts.

Slot area Aslot #1: 72.6 mm2

#2: 24.8 mm2

#3: 281 mm2

#4: 281 mm2

The area of a single slot.

Stamping area Acoil mm2 The area, including scrap, of the

electric steel required to stamp one stator and/or rotor.

Magnet area Amagnet 590 mm2 The total area of the magnet slots

in the rotor. The area is equal for all concepts.

Number of slots Nslots #1: 48

#2: 48 #3: 12 #4: 12

The total number of slots in a stator.

Slot circumference Cslot #1: 41.7 mm

#2: 23 mm

The circumference of one slot.

Copper fill factor F #1: 0.4 #2: 0.63 #3: 0.54

(43)

#4: 0.54

Density electric steel Des 7.65 kg/dm3

Density copper Dcu 8.96 kg/dm3

Density magnets Dm 7.5 kg/dm3

Stamping speed vstamping spm Sheets per minute that one press

line can produce.

Stamping capacity Nsheets sph Sheets per hit that one press line

can produce.

Annual production Nunits 1000000 Units/year Unit = Stator + Rotor

Work hours Twork 6000 h The total number of work hours

per year.

Machine availability U % The percentage of the uptime in relation to the total time of manufacturing

High volume production of Electric Machines : Results from WP3 in FFI project P43352-1