IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 5, SEPTEMBER/OCTOBER 2013 2053

Design and Optimization of A Blood Pump forA Wearable Artificial Kidney Device

Miroslav Markovic, Michael Rapin, Marc Correvon, and Yves Perriard

Abstract—The aim of the European project Nephron+ is thedesign of a wearable artificial kidney device. This paper is focusedon the design of the corresponding ultralow-hemolysis continuous-operation blood pump. Accurate specifications and operatingprinciple of the pump are determined. A first nonoptimal configu-ration of a linear electromechanical actuator which will be used topump the blood is designed. Its prototype is presented along withthe corresponding driving electronic circuit. Finally, based on themeasurements, the actuator is optimized, and the final design andfirst experimental results are presented.

Index Terms—Actuators, efficiency, hemolysis, medical controlsystems, microcontroller, pumps.

I. INTRODUCTION

THE kidneys serve the body as a natural filter of the blood,as they remove wastes which are diverted to the urinary

bladder [1]. In producing urine, the kidneys excrete wastessuch as urea and ammonium, and they are also responsible forthe absorption of water, glucose, and amino acids, keeping thebody’s internal equilibrium of water and minerals.

A. Dialysis

In the case of kidney disease or failure (which is referredto as nephropathy), an artificial replacement should be made,which means that an external device should remove waste fromthe blood. This artificial filtering process is called dialysis. Itis regarded as a “holding measure” until a renal transplant canbe performed (or, sometimes, as the only supportive measurein patients for whom a transplant would be inappropriate).However, this “holding” function should often last for a longtime, as the need for a kidney is much higher than the offer: InSwitzerland in 2009, the mean waiting time was 700 days [2].

Manuscript received October 2, 2012; revised November 22, 2012; acceptedJanuary 3, 2013. Date of publication May 3, 2013; date of current versionSeptember 16, 2013. Paper 2012-EMC-509.R1, presented at the 2012 IEEEEnergy Conversion Congress and Exposition, Raleigh, NC, USA, September15–20, and approved for publication in the IEEE TRANSACTIONS ON INDUS-TRY APPLICATIONS by the Electric Machines Committee of the IEEE IndustryApplications Society.

M. Markovic and Y. Perriard are with the Integrated Actuators Labora-tory (LAI), School of Engineering (STI), Ecole Polytechnique Federale deLausanne, 1015 Lausanne, Switzerland (e-mail: yves.perriard@epfl.ch).

M. Rapin and M. Correvon are with the Swiss Center for Electronics andMicrotechnics (CSEM), 2002 Neuchatel, Switzerland (e-mail: michael.rapin@csem.ch; marc.correvon@csem.ch).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TIA.2013.2260851

The main type of the dialysis is the hemodialysis. It con-sists in removing the waste by circulating blood outside thebody through an external filter, called a dialyzer, that containsa semipermeable membrane. The patient’s blood is pumpedthrough the blood compartment of a dialyzer. The blood flowsin one direction through the membrane, and a special liquid,called dialysate, flows in the opposite direction. The countercur-rent flow of the blood and dialysate maximizes the concentra-tion gradient of solutes between the blood and dialysate, whichhelps to remove more waste from the blood.

The main problem with dialysis is that the patient needs tobe placed in the hospital and that it is typically performed threetimes per week, with 4 h for each treatment. In addition, eachtreatment requires 120 L of dialysate. All this is, of course,related to a huge cost of treatment (typically, in Switzerland,80 000 euros per patient per year).

B. Envisaged Artificial Kidney

In order to significantly improve the patient’s quality of lifeand reduce the costs, the goal of the European project Nephron+[3], which started in April 2010 and will last for four years, is todesign a wearable device to replace the bulky dialyzer. Its mainproperty will be the continuous blood filtering, which is muchmore beneficial for the patient’s health, with a significantlyreduced quantity of dialysate. The participants in the projectare, among many others, the Swiss Center for Electronics andMicrotechnics (CSEM) from Neuchatel, Switzerland, and theEcole Polytechnique Federale de Lausanne, Switzerland.

The device is to be supplied from batteries and carried bythe patient. Hence, the device can be regarded as a wearableartificial kidney. This will be the first device of this type at themarket.

C. State of the Art

Many different studies have been done on wearable or im-plantable rotary blood pumps for ventricular assistance devices,which may be either centrifugal [4], [5] or axial [6]–[8].They show that the required blood flow and pressure drop aredifferent from the requirement of a blood pump needed fora hemodialysis system. Indeed, the blood flow for a kidneydevice is usually more than 30 times lower. Also, the requiredpressure in standard utilization is two times higher. Due to theseimportant differences, the mentioned rotary pump does not suitfor our use. Even if the rotary blood pump may reach therequired pressure drop for an artificial kidney, it would require a

0093-9994 © 2013 IEEE

2054 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 5, SEPTEMBER/OCTOBER 2013

Fig. 1. Principle of operation of the artificial kidney device.

higher rotational speed, which would significantly increase thehemolysis rate.

Lee et al. studied a valveless pulsatile blood pump [9]: Itis shown that this kind of pump provides better results thanconventional units in terms of hemolysis, which is an extremelyimportant characteristic in continuous extracorporeal bloodcirculation.

Other candidates for the pump are also analyzed. The elec-troosmotic pumps can generate the flow rate which is too smallfor the presented application, and in addition, they require a toohigh voltage (typically 100 V). The syringe-type pumps maygenerate clotting problems.

Finally, the references which served as inspiration for thepresented project are [10]–[14]. They present various config-urations of pumps, for blood and other liquids. Reference [15]presents the design of a pump for a portable renal replacementsystem. The principle is different from that which will bepresented in this paper: Several cams sequentially compressfingers, which compress flexible tubes, thus eliminating valves.The pump volume of 150 cm3 and the achieved flow rate of100 mL/min are similar to the performance which will bepresented in this paper.

II. ARTIFICIAL KIDNEY AND ITS BLOOD PUMP

A. Principle of Operation

The principle of operation of the artificial kidney device isshown in Fig. 1. It is a hemodialysis machine, in which thepatient’s blood is continuously, using a blood pump, circu-lated through a semipermeable extracorporeal membrane andreturned to the patient (blood circuit in red in Fig. 1).

The opposite side of the same membrane is washed withan electrolyte solution (dialysate) containing the normal con-stituents of plasma water. Diffusion (osmotic pressure) and con-vection (hydrostatic pressure) are the two mechanisms causinga flow of water and dissolved substances from the blood to thedialysate compartment.

A special pump is used to circulate the dialysate (dialysatecircuit in blue in Fig. 1). The undesired molecules are removedfrom the dialysate using a filter with nanoparticles in a sorbentunit.

The aim of this paper is to show the procedure of the designof the blood pump. For cost reasons, the pump must be reusablewithout excessive maintenance.

B. Blood Pump Specifications

The mean blood pressures in arteries and in veins are Pa =100 mmHg and Pv = 40 mmHg, respectively (750 mmHg =100 kPa). In addition, the semipermeable membrane will beopposed to the blood flow, which means that it will gener-ate a pressure drop. The corresponding value is estimated toPm = 450 mmHg. Hence, the blood pump should compen-sate the total pressure drop, which is therefore estimated toPp = Pm + Pv − Pa = 390 mmHg = 52 kPa. Note that thepositive difference Pa − Pv will “help” the pump during itsoperation.

The maximal necessary flow rate of the blood is estimated to200 mL/min, which means that Qp = 3.3× 10−6 m3/s. Thenecessary mechanical power to be provided by the pump ishence PpQp = 0.172 W. If we suppose the efficiency of 10%,the necessary electrical power to be consumed by the pump isapproximately 1.7 W. To complete the specifications, the pumpvolume and mass should not overpass 140 cm3 and 400 g,respectively.

C. Blood Pump Additional Requirements

Apart from the “mechanical” specifications concerning thepumping of blood, some other considerations should be takendue to the specificity of the blood. At first, blood is a compli-cated fluid. It contains red blood cells (RBCs) which containthe hemoglobin.

The RBC can be easily destroyed by one of the follow-ing three factors: mechanical force, sudden variation of thepressure, or the temperature. Concerning the first two factors,their critical values are not easy to estimate. Concerning thetemperature, it should not overpass 45 ◦C [16].

The process of RBC destruction is called hemolysis: Whenan RBC is destroyed, the hemoglobin is freed. It means thatthe increased concentration of hemoglobin means the increasedhemolysis. The normalized index of hemolysis (NIH) is aclinical measure of hemolysis measured as the concentration ofhemoglobin in the blood. Its value should not overpass 100 mgper 100 L of pumped blood.

In addition, if the blood stays immovable during a shorttime, there is a risk of thrombosis (formation of blood clot).Hence, the geometry of the pump should be made takingthis fact into consideration: It should not contain “corners” inwhich the blood can stay without moving. All the mentionedrequirements are too difficult to take into account during thepump design. Only the measurements on the pump prototypewill give responses if those requirements are met.

Finally, the pump should be “bloodtight” which means thatno blood can escape and (equally important) that the air cannotenter. An air bubble can have catastrophic consequences for thepatient [17].

MARKOVIC et al.: DESIGN AND OPTIMIZATION OF A BLOOD PUMP FOR A WEARABLE ARTIFICIAL KIDNEY DEVICE 2055

Fig. 2. Pump configuration and its operation in four steps [13].

III. PUMP CONFIGURATION AND OPERATION

A. Pump Configuration

After a preliminary study, it is decided that an optimal pumpconfiguration for this application is a linear peristaltic pumpwith two elastic tubes, as shown in Fig. 2. As it is peristaltic,it enables to know the blood flow rate without a sensor whichsignificantly improves the system reliability and cost.

The pump contains four actuators (A, B, C, and D) whichmove in the vertical direction. Using two tubes, when theactuator is to switch from one to another end (extreme) verticalpositions, the elastic force of the closed tube will help themovement toward the position where another tube should beclosed.

It is also important to point out that the actuator couples Aand C, and B and D, operate simultaneously and in the oppositedirection: When one from the couple goes up, another goesdown and vice versa.

In the closed state, the tube opening will be zero (tubecompletely pinched). In the open state, the tubes will not be leftcompletely without external force: A remaining force will keepthe tube opening equal to 50% of its diameter in the free state.This will help the tube to restore its shape during the refillingat the inlet of the pump, making the pump less sensitive to inletpressure.

B. Actuator Specifications

The chosen tube is made of silicone elastomer and producedby Maagtechnic. Its internal diameter is 8 mm; the wall thick-ness is 1 mm. This industrial tube does not fulfill the medicalnorms (at first hemocompatibility), but it will be used for thepump first prototype. The final prototype is intended to be madeusing the tubes made of Tygon 3350 by Saint-Gobain (whichfulfills the norms, more precisely ISO 10993 guidelines forcontact with blood), but due to nonstandard dimension, it hasa delivery time too long to wait for the first prototype.

The elastic force of the chosen tube is experimentally mea-sured. Finally, the necessary profile of the force which oneactuator should generate is given in Fig. 3. The position values

Fig. 3. Necessary force that should be generated by one actuator.

of +2/0/− 2 mm correspond to high-end/mean/low-end posi-tions, respectively.

The force is shown for the actuators B and C from Fig. 2,as they should generate higher force than A and D. Indeed,suppose that the actuator is in the low-end position and startsits moving toward the high-end position (such as B betweensteps 3 and 4). It is the low pressure (and force Flp) in the lowertube which helps the movement and the high pressure (and forceFhp) in the higher tube which is opposed to the movement.Therefore, the difference F = Fhp − Flp is shown in Fig. 3 asthe force F which should be generated by the actuator over thestroke of 4 mm.

After the calculations which take into account the tube di-mensions and the necessary blood flow rate, it is obtained thateach actuator should press the tube at its length of 12 mm withthe frequency of 2 Hz (so the sequence in Fig. 2 repeats twiceeach second).

IV. ACTUATOR PRELIMINARY DESIGN

A. Actuator Configuration

Using electromagnetic actuators is the best compromisebetween simplicity of construction, power consumption, andmaintenance [14]. Four candidates (1, 2, 3, and 4) are taken intoaccount for the actuator configuration (Figs. 4 and 5). Each ofthem is analyzed using a simple analytical model and a finite-element model (FEM) commercial software package.

After a quick preliminary analysis (the results will not bepresented here), the configurations 1 and 2 are eliminated, astheir constant force profile along the stroke does not correspondto the force requirement for the actuator (Fig. 3). Thus, in orderto generate a significant force in extreme positions, those actu-ators should have a huge volume (typically twice the volumerequired by specifications). The configurations 3 and 4 have anadvantage of having an important reluctant force at the end ofthe stroke, which allows maintaining the tube closed withoutcurrent.

Finally, after the first calculations, it turned out that theconfiguration 3 is not suitable, as its two windings require

2056 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 5, SEPTEMBER/OCTOBER 2013

Fig. 4. Actuator configurations (left) 1 and (right) 2.

Fig. 5. Actuator configurations (left) 3 and (right) 4.

significant volume and mass. Hence, the configuration 4 ischosen, which contains one winding.

B. Actuator Modeling

The actuator geometry with all its design parameters isshown in Fig. 6. The electromagnetic analytical model is basedon the analysis of the equivalent magnetic circuit representedin Fig. 7. The force is determined by the derivation of magneticenergy Wmag by using the formula Fy = −dWmag/dy, with yas the mover vertical position. The iron is assumed to be linear;this assumption is provided by the optimization process as itwill be shown later: The magnetic flux density is limited so thatthe iron stays far from the saturation.

In Fig. 7, ΛIron1, ΛIron2, and ΛIron3 represent the reluc-tances of the iron; Λi is the internal reluctance of the magnets;Λh and Λl represent the (variable) reluctances of the higherand the lower air gaps, respectively; Θm and Θcoil are themagnetic potentials (magnetomotive forces) of the magnets andthe coil, respectively; and Φmh, Φml, and Φcoil are the resultingmagnetic fluxes flowing in the magnetic circuit branches.

All of the previous reluctances are calculated using the geo-metrical parameters of the actuator and the magnetic properties

Fig. 6. Chosen actuator configuration with its geometrical parameters.

Fig. 7. Equivalent magnetic circuit of the actuator.

of the iron and the magnet. The flux leakage in the air gap,represented in Fig. 8, is neglected in the presented model, andonly the reluctance enclosed by the red rectangle is taken intoaccount. Again, the reluctances of the air gaps highly dependon the vertical position (y) of the mover.

Finally, the magnetic energy Wmag is calculated as a functionof the mover position. The energy is calculated in each partof the magnetic circuit (magnets, air gaps, and iron yokes), byknowing that the energy volume density is given by wmag =(BH)/2. It is worth noting that the main part of the energy iscontained in the air gaps and in the magnets. Finally, all thoseenergy components are added, and the total energy is derived toobtain the force as already explained.

MARKOVIC et al.: DESIGN AND OPTIMIZATION OF A BLOOD PUMP FOR A WEARABLE ARTIFICIAL KIDNEY DEVICE 2057

Fig. 8. Simulated magnetic flux lines in the air gap.

Fig. 9. First actuator prototype.

Fig. 10. First actuator prototype with two tubes.

C. Actuator First Prototype

The first actuator prototype is not a result of optimization.Instead, the values of its parameters are quickly chosen tomore or less satisfy the specifications, with the goals to verifythe actuator electromagnetic analytical model, to generate theactuator thermal analytical model, and, finally, to refine thespecifications. Based on all this, the second prototype will bethe result of a detailed optimization.

The first actuator prototype is shown in Figs. 9 and 10. Itsmass is 124 g.

Fig. 11. Force generated by the actuator: Comparison between the analyticalmodel, FEM, and experimental results.

Fig. 12. Actuator temperature measured by a thermal camera (the ambienttemperature is 22.2 ◦C).

D. Measurements on the Actuator First Prototype

In order to verify the electromagnetic analytical model, theforce is measured. A constant current of 1.1 A is injected inthe winding. In addition, the actuator is simulated using a FEMcommercial software which calculates the force. Fig. 11 showsthe corresponding results and confirms that the electromagneticanalytical model is correct. The differences between the resultsfor the position above 1 mm are due to the saturation of theferromagnetic parts.

In order to be able to generate a simple thermal analyticalmodel, a constant power of 2.92 W is dissipated in the winding,and the thermal steady state is established. Fig. 12 shows thetemperature measured using a thermal camera.

Obviously, the whole mover can be regarded as an isothermalbody. Using the measured temperatures and dissipated power,the thermal resistance between the mover and surrounding airis obtained as 23.1 K/W. By supposing that this resistancecorresponds to the natural convection over the mover surface,the corresponding convection coefficient is 14.7 W/(m2 · K).This value will be taken to calculate the thermal resistance inthe simple thermal model which will be used to optimize theactuator.

It is important to note here that only the winding surfacetemperature will be calculated in the model. This value has to

2058 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 5, SEPTEMBER/OCTOBER 2013

Fig. 13. Power card.

be limited to 45 ◦C according to the specifications. The windinginsulation will not be a critical element in this case as it cansupport the temperature of 80 ◦C.

V. CONTROL ELECTRONIC CIRCUITS

The control system for the operation of the blood pump iscomposed of a central computer, four power cards, and fourposition measurement cards (one per actuator). The centralcomputer communicates with the power cards via the portRS485.

The power card (Fig. 13) contains a microcontrollerSTM32F103 which gives the ON/OFF commands for the powertransistors according to the pulsewidth-modulation logic. Theactuator is driven by the MC33887 integrated H-bridge. Also,each power card receives a signal from the position measure-ment card.

The position is determined by measuring two capacitances:between the upper yoke and mobile yoke and between thelower yoke and mobile yoke. In order to do this, the positionmeasurement card sends two rectangular voltage signals to theupper and lower mobile yokes and detects the voltage of themobile yoke. After demodulation, the resulting analog signal,which is proportional to the position, is sent to the power card.

VI. ACTUATOR OPTIMAL DESIGN

The final goal is to obtain the optimal actuator configurationusing the verified analytical models. The optimization will beperformed using the software package ProDesign [18] and isinspired by [19].

A. Required Force Profile

The actuator force has to satisfy the required profile (Fig. 3),which means that this profile will impose some constraintsfor the optimization. As the optimization tool is not able toaccept this profile as one constraint, it is decided to “sample”the profile in eight points. Those points (values of position and

TABLE IPOINTS FROM THE FORCE PROFILE

TABLE IIOPTIMAL ACTUATOR CONFIGURATION

TABLE IIIOPTIMAL ACTUATOR MAIN PARAMETERS

force) are presented in Table I. The force corresponding to thepoint 8 has to be achieved without current, i.e., only by thereluctant force.

B. Optimization

In order to perform the optimization, the software ProDesignneeds at first the actuator analytical model. Concerning theconstraints, eight points from Table I impose eight constraints.

In addition, the flux density in the iron is limited to 1.5 T; thetemperature is limited to 45 ◦C as required by specifications. Asthe specifications impose 400 g as the maximal pump mass, itmeans that the maximal actuator mass is limited to 100 g. Thevalue of the current density is left free, in order to be adapted toprovide the necessary force for each point.

Therefore, a complete theoretical model including electro-magnetic force calculation in eight points and a thermal modelis introduced in the optimization software; the chosen optimiza-tion function is simply the volume of the actuator. Thus, theoptimization software varies all the parameters simultaneouslyto minimize the volume accordingly to the defined constraints.To perform the optimization, the software applies an advancedsequential quadratic programming algorithm, which is an itera-tive method for nonlinear optimization.

Finally, Tables II and III present the optimal actuator con-figuration. The iron is Armco; the magnet remanence is 1.4 T.The optimal actuator drawing is shown in Fig. 14. The set offour actuators is presented in Fig. 15, and the whole system ispresented in Fig. 16.

MARKOVIC et al.: DESIGN AND OPTIMIZATION OF A BLOOD PUMP FOR A WEARABLE ARTIFICIAL KIDNEY DEVICE 2059

Fig. 14. Optimal actuator.

Fig. 15. Set of four actuators.

Fig. 16. Whole system.

VII. EXPERIMENTAL RESULTS

The very first experiments on the optimal actuator showpromising results. Fig. 17 shows the necessary actuator force(imposed by the specifications) compared to the simulation andexperimental results and the corresponding current injected inthe coil. All the three forces match well.

Concerning the operation of the whole pump, the first ex-periments are performed using a constant current of 1.2 A(maximal value from Fig. 17). In that case, the pump generatesthe flow rate of 150 mL/min (0.75 L is pumped in 5 min) andthe pressure drop of 38 kPa.

The required flow rate of 200 mL/min will be achieved byincreasing the frequency above 2 Hz, and the required pressureof 52 kPa will be achieved by increasing the current.

Fig. 17. Optimal actuator force.

VIII. SUMMARY AND FUTURE WORK

This paper has presented the design of a blood pump for awearable artificial kidney device. The first experiments haveshown that the pump satisfies the desired specifications in termsof rate of flow and pressure drop. The prototype is now ready forthe next steps: measurement of NIH in a specialized laboratory.

The project Nephron+ continues until April 2014. A firstdemonstrator containing the real-time microfluidic circuit boardintegrating the physical sensors, some actuators, and severalfiltering systems is now available [3]. The NIH measurementtests of the blood pump will start in February 2013.

The main drawback of the device is the noise which itgenerates. In the actual state, it cannot be used, as the noisewould be annoying for the patient. In order to address thisproblem, the control algorithm will be modified for the actuatorfinal version, so as to reduce the speed at closing of the tubes.

REFERENCES

[1] Internet encyclopedia, Aug. 5, 2013. [Online]. Available: http://en.wikipedia.org/wiki/Kidney

[2] Swiss Confederation website, Aug. 5, 2013. [Online]. Available: www.bag.admin.ch

[3] Official website of the European project, Aug. 5, 2013. [Online]. Avail-able: www.nephronplus.eu

[4] M. Ertan Taskin, K. H. Fraser, T. Zhang, B. Gellman, A. Fleischli,K. A. Dasse, B. P. Griffith, and Z. J. Wu, “Computational characterizationof flow and hemolytic performance of the ultramag blood pump for circu-latory support,” Artif. Org., vol. 34, no. 12, pp. 1099–1113, Dec. 2010.

[5] W. Hijikata, H. Sobajima, T. Shinshi, Y. Nagamine, S. Wada, S. Takatani,and A. Shimokohbe, “Disposable maglev centrifugal blood pump utiliz-ing a cone-shaped impeller,” Artif. Org., vol. 34, no. 8, pp. 669–677,Aug. 2010.

[6] S. Cheng, M. W. Olles, D. B. Olsen, L. D. Joyce, and S. W. Day, “Minia-turization of a magnetically levitated axial flow blood pump,” Artif. Org.,vol. 34, no. 10, pp. 807–815, Oct. 2010.

[7] S.-M. Yang and C.-C. Lin, “Performance of a single-axis controlled mag-netic bearing for axial blood pump,” in Conf. Rec. 42nd IEEE IAS Annu.Meeting, 2007, pp. 963–968.

[8] F. Wang, J. Wang, Z. Kong, and L. Xu, “A novel BLDC motor with passivemagnetic bearings for blood pump application,” in Conf. Rec. 38th IEEEIAS Annu. Meeting, 2003, pp. 1429–1433.

[9] K. Lee, C. H. Mun, S. R. Lee, B. G. Min, K. J. Yoo, Y. W. Park, andY. S. Won, “Hemodialysis using a valveless pulsatile blood pump,”ASAIO J., vol. 54, no. 2, pp. 191–196, Mar./Apr. 2008.

2060 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 49, NO. 5, SEPTEMBER/OCTOBER 2013

[10] S. Kim, S. Hashi, and K. Ishiyama, “Actuation of novel blood pump bydirect application of rotating magnetic field,” IEEE Trans. Magn., vol. 48,no. 5, pp. 1869–1874, May 2012.

[11] S. Yang and M. Huang, “Design and implementation of a magneticallylevitated single-axis controlled axial blood pump,” IEEE Trans. Ind.Electron., vol. 56, no. 6, pp. 2213–2219, Jun. 2009.

[12] E. Lim, S. Dokos, S. L. Cloherty, R. F. Salamonsen, D. G. Mason,J. A. Reizes, and N. H. Lovell, “Parameter-optimized model ofcardiovascular–rotary blood pump interactions,” IEEE Trans. Biomed.Eng., vol. 57, no. 2, pp. 254–266, Feb. 2010.

[13] M. Hu, H. Du, and S. Ling, “A digital miniature pump for medical ap-plication,” IEEE/ASME Trans. Mechatronics, vol. 7, no. 4, pp. 519–523,Dec. 2002.

[14] A. G. S. Barreto Neto, A. M. N. Lima, H. Neff, C. L. Gomes,and C. Moreira, “Linear peristaltic pump driven by three magneticactuators: Simulation and experimental results,” in Proc. IEEE I2MTC,2011, pp. 1–6.

[15] J. Kang, S. Tamera, J. D. Weaver, D. N. Ku, and D. W. Rosenc, “Pumpdesign for a portable renal replacement system,” J. Med. Dev., vol. 5, no. 3,pp. 031008-1–031008-8, Sep. 2011.

[16] J. Yeun and T. Depner, Principles of Hemodialysis. Philadelphia, PA,USA: Elsevier, 2005.

[17] A. Davenport, V. Gura, C. Ronco, M. Beizai, C. Ezon, and E. Rambod, “Awearable haemodialysis device for patients with end-stage renal failure:A pilot study,” Lancet, vol. 370, no. 9604, pp. 2005–2010, Dec. 2007.

[18] [Online]. Available: designprocessing.com[19] J. Ji, W. Zhao, G. Liu, and F. Wang, “High reliability linear drive

device for artificial hearts,” J. Appl. Phys., vol. 111, no. 7, pp. 07E729-1–07E729-3, Apr. 2012.

Miroslav Markovic was born in Arandjelovac,Serbia, in 1970. He received the B.S. degree fromthe Faculty of Electrical Engineering, University ofBelgrade, Belgrade, Serbia, in 1996 and the Ph.D.degree from the Ecole Polytechnique Federale deLausanne (EPFL), Lausanne, Switzerland, in 2004.

He is currently a Project Leader with the In-tegrated Actuators Laboratory (LAI), School ofEngineering (STI), EPFL. His research interest isoptimization design of high-performance electricdrives.

Michael Rapin was born in Payerne, Switzerland,in 1985. He received the B.S. degree from theHaute Ecole d’Ingénierie et de Gestion du Cantonde Vaud, Yverdon-les-Bains, Switzerland, and theM.S. degree in electrical engineering from the EcolePolytechnique Federale de Lausanne, Lausanne,Switzerland.

He is currently an R&D Engineer with the SwissCenter for Electronics and Microtechnics (CSEM),Neuchatel.

Marc Correvon was born in Neuchatel, Switzerland,in 1956. He received the M.S. degree from the EcolePolytechnique Federale de Lausanne, Lausanne,Switzerland.

He was the Project and System Manager as wellas the Head of Electronics with ETEL Space andAlcatel Space Switzerland. For nine years, he hasbeen a Professor with the University of AppliedSciences Western Switzerland, Yverdon-les-Bains,Switzerland. In 2008, he joined the Swiss Center forElectronics and Microtechnics (CSEM), Neuchatel,

as the Section Head of electronics and firmware in the Systems EngineeringDivision. His research activities include embedded electronics and preprocess-ing of physiological parameters as well as medical device development.

Yves Perriard was born in Lausanne, Switzerland,in 1965. He received the M.Sc. degree in microengi-neering and the Ph.D. degree from the Ecole Poly-technique Federale de Lausanne (EPFL), Lausanne,Switzerland, in 1989 and 1992, respectively.

He was a Cofounder and Chief Executive Offi-cer of Micro-Beam SA, which is involved in high-precision electric drive design. He is currently theDirector of the Laboratory of Integrated Actuators(LAI), School of Engineering (STI), EPFL, wherehe was a Senior Lecturer beginning in 1998 and has

been a Professor since 2003. His research interests are in the field of newactuator design and associated electronic devices.

Dr. Perriard has been the Vice-Director of the Microengineering Institute inNeuchatel since 2009.