Design and Prototyping of a Low-cost Portable Mechanical Ventilatorby

Stephen K. Powelson

SUBMITTED TO THE DEPARETMENT OF MECHANICAL ENGINEERING IN PARTIAL

FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

BACHELOR OF SCIENCE IN MECHANICAL ENGINEERINGAT THE

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

JUNE 2010

C 2010 Stephen K. Powelson. All rights reserved.

The author hereby grants to MIT permission to reproduce

and to distribute publicly paper and electronic

copies of this thesis document in whole or in part

in any medium now known or hereafter created.

r

ARCHIVES

MASACHSETS NS EOF TECHNOLOGY

JUN 30 2010

LIBRARIES

fNl

Signature of Author:Department of Mechanical Engineering

May 23, 2010

Certified by:Alexander H. Slocum

Pappalardo Professor of Mechanical EngineeringThesis Supervisor

Accepted by:4:' John H. Lienhard V

ofessor of Mechanical Engineering

Chairman, Undergraduate Thesis Committee

Design and Prototyping of a Low-cost Portable Mechanical Ventilator

Abdul Mohsen Al Husseini', Justin Negretel, Stephen Powelson', Amelia Servi', Alexander Slocum',Jussi Saukkonen2

'Massachusetts Institute of Technology, Department of Mechanical Engineering2Boston University, School of Medicine

AbstractThis paper describes the design and prototyping of a low-cost portable mechanical ventilator foruse in mass casualty cases and resource-poor environments. The ventilator delivers breaths bycompressing a conventional bag-valve mask (BVM) with a pivoting cam arm, eliminating theneed for a human operator for the BVM. An initial prototype was built out of acrylic, measuring11.25 x 6.7 x 8 inches (285 x 170 x 200 mm) and weighing 9 lbs (4.1 kg). It is driven by astepper motor powered by a 14.8 VDC battery and features an adjustable tidal volume of up to900 mL, adjustable breaths per minute (bpm) of 5-30, and inhalation to exhalation time ratio (i:eratio) options of 1:2, 1:3 and 1:4. Tidal volume, breaths per minute and i:e ratio are set via user-friendly knobs, and the settings are displayed on an LCD screen. The prototype also features anassist-control mode and an alarm to indicate over-pressurization of the system. Future iterationsof the device will be fully calibrated to medical standards and include all desired ventilatorfeatures. Future iterations will be further optimised for low power-consumption and will bedesigned for manufacture and assembly. With a prototyping cost of only $420, the bulk-manufacturing price for the ventilator is estimated to be less than $100. Through this prototype,the strategy of cam-actuated BVM compression is proven to be a viable option to achieve low-cost, low-power portable ventilator technology that provides essential ventilator features at afraction of the cost of existing technology.

Keywords: Ventilator, Bag Valve Mask (BVM), Low-Cost, Low-Power, Portable andAutomatic

1. Introduction

Respiratory diseases and injury-inducedrespiratory failure constitute a major publichealth problem in both developed and lessdeveloped countries. Asthma, chronicobstructive pulmonary disease and other chronicrespiratory conditions are widespread. Theseconditions are exacerbated by air pollution,smoking, and burning of biomass for fuel, all ofwhich are on the rise in developing countries" 2

Patients with underlying lung disease maydevelop respiratory failure under a variety ofchallenges and can be supported mechanical

ventilation. These are machines thatmechanically assist inspiration and exhalation, a

3process referred to as artificial respirationWhile the ventilators used in modem hospitals inthe United States are highly functionally andtechnologically sophisticated, their acquisitioncosts are correspondingly high (as much as$30,000). High costs render suchtechnologically sophisticated mechanicalventilators prohibitively expensive for use inresource-poor countries. Additionally, theseventilators are often high-maintenance, requiringcostly service contracts from themanufacturer. In developing countries, this has

My contributions to the low-cost ventilator project were principally in the areas ofmechanical design and construction.

The project was started as a part of the class 2.75: Precision Machine Design. In the

beginning phase of the project, during group brainstorming we decided to actuate a bag-valve-mask to make our low-cost ventilator. While considering different actuation mechanisms Isuggested making a rolling-contact cam; as a group we then worked out the specifics of this cammechanism. Once this cam strategy was decided, I was responsible for the Solidworks modelingand much of the construction of the first cam-compression proof-of-concept rig. After the rigshowed that cam compression was viable, I created a Solidworks model of the first prototype.We constructed this first prototype during the latter half of the fall semester. I was responsiblefor much of the mechanical construction while other group members were responsible forelectronics and programming of the device.

During the spring semester, 2.75 was continued as the class 2.752. During this semester,we sought to improve our design to build a fully-functioning prototype. After muchbrainstorming of driving mechanisms, we opted to use a double-ended stepper motor drivinggeared cams as our actuation mechanism. I was responsible for the Solidworks modeling of thisnew prototype, and later responsible for the majority of the mechanical construction. Once theprototype was built, we realized the cam needed to be optimized for maximum compression.After exploring several ideas, I made necessary modifications and tested the prototypesperformance with a bench-level experiment.

I was additionally responsible for all of the Solidworks modeling of the DFMA conceptof the ventilator. This model is still a work-in-progress. Finally, I was responsible for writingmost of the mechanical design sections of our final paper.

led to practices such as sharing ventilatorsamong hospitals and purchasing less reliablerefurbished units. Since medical resources inthese countries are concentrated in major urbancenters, in some cases rural and outlying areashave no access at all to mechanical ventilators.The need for an inexpensive transport ventilatoris therefore paramount.

In the developed world, where well-stockedmedical centers are widely available, theproblem is of a different nature. While there areenough ventilators for regular use, there is a lackof preparedness for cases of mass casualty suchas influenza pandemics, natural disasters andmassive toxic chemical releases. The costs ofstockpiling and deployment of state-of-the-artmechanical ventilators for mass casualty settingsin developed countries are prohibitive.According to the national preparedness planissued by President Bush in November 2005, theUnited States would need as many as 742,500ventilators in a worst-case pandemic. Whencompared to the 100,000 presently in use, it isclear that the system is lacking4.

One example of this shortage occurredduring Hurricane Katrina, when there wereinsufficient numbers of ventilators 5 , andpersonnel were forced to resort to manual BVMventilation. Measures to improve preparednesshave since been enacted. Most notably theCenter for Disease Control and Prevention(CDC) recently purchased 4,500 portableemergency ventilators for the strategic nationalstockpile 7 . However, considering the lownumber of stocked ventilators and their currentlyhigh cost, there is a need for an inexpensiveportable ventilator for which production can bescaled up on demand.

Prior Art

While many emergency and portableventilators are on the market, an adequate low-cost ventilator is lacking. A cost-performancedistribution is depicted in Figure 1 withmanually operated BVMs on the low end of cost

and performance, and full-featured hospitalventilators on the other extreme. The middlesection of the chart includes the existing portableventilators which can be broadly categorized aspneumatic and electric. Pneumatic ventilatorsare actuated using the energy of compressed gas,often a standard 50 psi (345 kPa) pressuresource normally available in hospitals. Theseventilators have prices ranging from $700-1000.This category includes products such as theVORTRAN Automatic Resuscitator (VARTM), asingle patient, disposable resuscitator, and thereusable Lifesaving Systems Inc.'s Oxylator, 0-Two CAREvent@ Handheld Resuscitators andAmbu@ Matic. However, these systems cost anorder of magnitude more than our target priceand depend on external pressurized air, aresource to which our target market may nothave access.

0Kt6*

how"c*w

Figure 1: Cost performance distribution ofventilators

Electric ventilators are capable of operatinganywhere, and thus are not bound by thisconstraint. Ventilators of this type such asCareFusion LTV@ 1200 were the choice of theCDC for the Strategic National Stockpile. TheLTV@ 1200 weighs 13.9 lbs (6.3 kg) andincludes standard features as well as thecapability to slowly discontinue or wean offmechanical ventilator support. Its complexityelevates its cost to several thousands of dollars,an order of magnitude above our target retailprice.

................ ........ ... ... ..... .. I

The United States Department of Defencehas also developed several rugged, portableelectric ventilators. One such ventilator is theJohns Hopkins University (JHU) AppliedPhysics Laboratory (APL) Mini Ventilation Unit(JAMU), which weighs 6.6 lbs (3.0 kg),measures 220 cubic inches (3,600 cubic cm) andcan operate up to 30 minutes on a battery. Thisdevice was patented by JHU/APL, and licensedto AutoMedx. The commercial device featuressingle-knob operation. Its simplicity comes witha compromise, as the tidal volume, breath rateand other parameters cannot be adjusted by theuser, making it not suitable for many patientswho cannot tolerate the fixed tidal volume, rateor minute ventilation. It also cannot be operatedfor long periods in a resource-poor environment.Additionally, with a price tag over $2000, itcosts several times more than our target price.Another device, the FFLSS, weighs 26.5 lbs (12kg) and is capable of one hour of operationpowered by a battery. It also includes additionalphysiologic sensors, and fits in a standard U.S.Army backpack 8 . While these devices arefunctionally adequate, their compressors requirehigh power which limits battery life. In addition,their many pneumatic components are costly andare not easily repairable in a resource-poorenvironment.

Medical Device Requirements

In light of the aforementioned constraints, aset of mechanical, medical, economic, user-interface and repeatability functionalrequirements were developed. These include theASTM F920-93 standard requirements9, and aresummarized in Table 1.

Table 1: Device functional requirements

- User-specified breath/mininsp./exp ratio, tidal volume

- Assist control- Positive end-expiratory

Medical pressure (PEEP)- Maximum pressure limiting- Humidity exchange- Infection control- Limited dead-space- Portable- Standalone operation- Robust mechanical,

Mechanical electrical and softwaresystems

- Readily sourced andrepairable parts

- Minimal power requirement- Battery-powered

Economic - Low-cost (<$500)- Alarms for loss of power,

loss of breathing circuitUser- integrity, high airwayinterface pressure and low battery

life- Display of settings and

status- Standard connection ports- Indicators within 10% of

Repeatability correct reading- Breath frequency accurate

to one breath per minute

2. Device Design

Air Delivery Technique

Two main strategies were identified for theventilator's air delivery system. One strategyuses a constant pressure source to intermittentlydeliver air while the other delivers breaths bycompressing an air reservoir. The latter approachwas adopted as it eliminates the need for the

continuous operation of a positive pressuresource. This reduces power requirements andthe need for expensive and difficult to repairpneumatic components.

Whereas most emergency and portableventilators are designed with all custommechanical components, we chose to take anorthogonal approach - by building on theinexpensive BVM, an existing technology whichis the simplest embodiment of a volume-displacement ventilator. Due to the simplicity oftheir design and their production in largevolumes, BVMs are very inexpensive(approximately $10) and are frequently used inhospitals and ambulances. They are also readilyavailable in developing countries. Equipped withan air reservoir and a complete valve system,they inherently provide the basic needs requiredfor a ventilator.

The main drawback with BVMs is theirmanual operation requiring continuous operatorengagement to squeeze the bag. This operatingprocedure induces fatigue during long operations,and effectively limits the usefulness of thesebags to temporary relief. Moreover, an untrainedoperator can easily damage a patient's lungs byover compression of the bag. Our methodology,therefore, was to design a mechanical device toactuate the BVM. This approach results in aninexpensive machine providing the basicfunctionality required by mechanical ventilatorstandards.

Compression Mechanism

BVM compression can be achieved withlinear actuation (e.g. lead screw, rack and pinionor crankshaft), radial actuation (e.g. noose), orrotary actuation (e.g. cam). The linear actuationconcept was quickly rejected because thenecessary linear bearings add excessive cost,complexity and bulk to the actuator. The radialcompression concept was tested but found to beinefficient. This is because the bag has a high-friction exterior so a compression belt does noteasily slide along the surface. This can be

remedied with a roller chain in place of a belt.However, rollers introduce new problems suchas pinching. The bag is also not designed forradial actuation and such actuation producesinstabilities that result in non-linear compression.In contrast, radial actuation as achieved with acam mimics the hand motion for which the bagis designed. It is a compact design and employsrolling contact with the bag, thereby reducingfrictional losses between the actuator and thebag. This concept was thus pursued.

Cam Concept

The cam concept utilizes a crescent-shaped camto compress the BVM. This motion allowssmooth, repeatable deformation to ensureconstant air delivery (Figure 2). As it rotates, thecam makes rolling contact along the surface thusachieving high-efficiency power transmission.By controlling the angle of the cam, the amountof air volume delivered can be accuratelycontrolled. The cam mechanism was found to bethe most space and energy efficient of thecompression methods we considered.

Figure 2: Sketch of cam actuator

3. Prototype Design

First Prototype Design

The first prototype featured a DC motorwith a planetary gearbox (model PK51 fromrobotmarketplace.com) coupled to the cam. Theenclosure's frame consisted of four verticallymounted sections of 1/2" (12.5 mm) clear acrylic,each mounted to the bottom section of the frame

with interlocking mates. This prototype wasdesigned for use with an AmbuTM BVM. Thetwo inner sections (ribs) have U-shaped slotsmade to conform to BVM's surface contour,while the end sections feature openings toaccommodate the BVM's valve neck andoxygen reservoir.

Second Prototype Design

The DC motor used for the first prototypedid not deliver enough torque to achieveadequate volume delivery. Therefore, a secondprototype was constructed with the goal ofimproving performance with increased torqueoutput. Changes in the cam design and drivingmechanism were instrumental in improvingcompression performance. The cam assembly inthis prototype features two geared nylon camsdriven off of pinions mounted onto the shafts ofa double-shafted stepper motor (Figure 3).

Figure 3: Model of the cam assembly with motor

The radius of curvature of the camsapproximately matches that of the BVM, givinga mostly rolling-contact compression. The camarms are laser cut and their teeth are modulus 1with a pitch diameter of 96 mm. A double-endedstepper motor drives each of these cam armswith a 12 tooth pinion mounted on each end ofthe shaft. This gearing allows for a reduction of8 to 1. Centered between the two geared camarms is a rib which protrudes above the geared

cams on either side. This central rib makes firstcontact with the BVM. As the BVM is designedto be squeezed at the center of the bag, thiscentral rib greatly reduces the BVM's resistanceto compression.

This cam assembly was mounted in anacrylic compartment similar to that of the firstprototype. However a few notable changes tothe compartment were implemented. The secondprototype was designed for use with a LaerdalTMBVM. This BVM was selected because itfeatures an exhale valve and mask that can bedetached from the bag. This feature allows thevalve and mask to be placed at the patient end ofan extension tube. With the second prototype,the hinge location was shifted from the top edgeof the box to the center, in plane with the centeraxis of the BVM. This allows for better lidclosing and constraint of the BVM. Moreover,alterations were made to the central ribs thatconstrain the BVM in order to minimize frictionduring compression. The second prototype isshown in Figure 4.

Testing

A set of experiments were performed on thesecond prototype to determine volume deliveryand torque requirements. Flow rate wasmeasured using a spirometer connected to theoutput of the BVM. Integrating with respect totime yields volume delivery of eachcompression stroke. Figure 5 shows a plot of thevolume delivered with respect to cam angle. Theplot shows that over the range of typical tidal

.............. ......... - ... ...... .

volume settings, this relationship is linear. Thisrelation allows for precise control of volumedelivery.

0.6

1,445 55 65 75

CmAg(degee)Figure 5: Volume delivered versus cam angle

The force required to hold the cam at thepeak of its stroke was measured using a forcegauge. Using the known moment arm of the cam,these force measurements were then convertedto torque values. It was found that the necessaryholding torque is linearly related to volumedelivered. The maximum cam torque required tocompress the BVM was 2.64 N-m.

The torque output of the stepper motor isinversely related to the speed at which it isdriven. Therefore, at lower driving speeds thedevice is able to deliver a higher tidal volume.At 5 bpm with i:e ratio of 1:2, the maximumvolume delivery of the prototype is 0.9L. At 10,15 and 20 bpm and with the same i:e ratio, themaximum volume delivery drops off to 0.8L,0.7L, and 0.6 L respectively. While these resultsare not ideal, future iterations of the design willensure that for all combinations of user inputs,the desired tidal volumes can be delivered. Thisshould be achievable by further optimization ofthe shape of the cam and the constraints on thebag or increasing the gear ratio and motor size.

4. Control Implementation

Control design

The ventilator described in this paper canoperate in two modes: volume control (VC) Controller

I I . - "I 'll - 1111111111 11 ---- : . .............. .... .............

mode and assist control (AC) mode. Theventilator automatically operates in volumecontrol mode but can at any time be overriddeninto AC mode by a breath attempt by the patient.To operate the ventilator, the medical providerselects the tidal volume (typically 6-8 mL/kg ofideal body weight), breaths per minute and i:eratio. This provides a minimum assured minuteventilation (Ve). If at any time the patientspontaneously attempts a breath, a negativeinspiratory pressure that exceeds 2cmH20vacuum triggers the assist control mode and theventilator immediately delivers the set tidalvolume.

The advantage of the volume control modeis that the patient has an assured minuteventilation to ensure adequate gas exchange. Thepresence of the AC mode helps maintainsynchrony between the patient's breath patternand the breaths delivered by the ventilator.However, a disadvantage of the AC mode is thatif the patient is tachypneic, respiratory alkalosismay develop or, for those with obstructive lungdisease, air trapping may occur, raising intra-thoracic pressure with adverse hemodynamicand gas exchange consequences. These issuesare commonly handled with sedation. Thecombination of volume control and assist controlis adequate for the management of the majorityof clinical respiratory failure scenarios.

Parameters

The operator adjusts tidal volume, breathsper minute, and i:e ratio using three continuousanalog knobs mounted to the outside of theventilator. The prototype has a range of 0-900mL tidal volume, 5-30 breaths per minute (bpm)and 1:2 to 1:4 i:e ratio. However, these valuesdo not reflect the limits of the final design onlythe settings of the prototype. The ranges on thefinal design will be determined in consultationwith respiration specialists to allow for thebroadest range of safe settings.

An off-the-shelf Arduino Nanomicrocontroller board was selected to controlour device. The microcontroller runs a simplecontrol loop to achieve user-prescribedperformance. The control loop is triggered bythe internal timer set by user inputs, with theinspiratory stroke initiated at the beginning ofthe loop. Once the prescribed tidal volume isreached, the actuator returns the cam back to itsinitial position and holds until the next breath.The loop then repeats to deliver intermittentbreaths. If the loop is interrupted by a breathattempt (sensed through the pressure sensor), theventilator immediately delivers a breath,interrupting the loop and resetting the timer. Adiagram showing this loop is found in Figure 6.

over pressure detected ?

No Yes

V V

inhale attempt detected ?

Yes No:

V

begin breath cycle delay

+alarm

BMuser input

Volume, VE ratio

Figure 6: Ventilator control loop

Motor

According to initial experiments, amaximum torque of 2.64 Nm was required formaximum volume delivery of 0.9L. This valuewas used to size the motor chosen for theprototype. The stepper motor chosen is a bipolarconfiguration unit from Anaheim Automation,with part number 23Y002D. This is a squareNEMA 23 frame size motor with double outputshafts, which can provide 0.55 N-m of torque.With a gear ratio of 8:1, the cam can provide amaximum of 4.4 N-m of holding torque. Thehigher inductance coils (21.6 mH per phase) forthis motor are acceptable due to the low speedrange at which it will be operating.

Motor Driver

The motor driver comprises two H-Bridgecircuits. These circuits direct current through themotor in opposite directions, depending onwhich set of switches on the circuits areenergized. Speed of the motor is controlled bythe pulse train generated by the Arduino. Thepower is supplied directly from the battery, sothe only current limit is the current capability ofthe chip and battery.

We opted to use the Solarbotics@ motordriver, which is capable of supplying 2.92 ampsof current to one circuit. While compressing theBVM, the motor was measured to draw 1.5amps at a 50% duty cycle. This value is wellbelow the peak current capabilities of thecomponents.

User Interface

The three user inputs (tidal volume, bpm andi:e ratio) are set via three potentiometer knobs.An LCD displays the current settings, as well asan error if high airway pressure is detected.There is a main power switch that disconnectsthe battery when the device is not in use.

Safety Features

To ensure that the patient is not injured,patient airway pressure is monitored with apressure sensor connected to a sensor output onthe BVM. The same pressure sensor used forassist control also triggers an alarm if systempressure rises too high, alerting the physician toattend to the patient. As a further safety measureto prevent over-inflation, future iterations of thedevice will include a mechanical pressure reliefvalve.

Power Delivery

An AC/DC converter can be used to powerthe ventilator directly from a wall outlet or avehicle inverter. When external power isunavailable, the ventilator can run off of any

................................................................................ .

battery capable of delivering 12-15 volts and 3.5Amps. For the prototype, we used a 14.8 V,four-cell Li-Ion battery pack capable of 4.2Amps, with a capacity of 2200 mA-hr.

5. Conclusions

An improved working prototype of theventilator has been developed, relying on adouble-shafted stepper motor and cams withintegrated gear profiles to achieve BVMcompression. The prototype has user-controlledtidal volume, breaths per minute and inspiratoryto expiratory time ratio. It is also equipped witha pressure sensor allowing assist control and analarm indicating over-pressurization of thesystem. The device has low power requirements,with a time-averaged power consumption ofapproximately 5W. It is portable, weighing 16lbs (7.3 kg) and measuring 12 x 6.2 x 9 inches(305 x 157 x 229 mm), and features a handle andeasy to use latches. An LCD screen on the userpanel displays the user-prescribed settings. Apreliminary DFMA model has been designed,and further development is planned. A scaled-down pediatric version will also be considered.Extensive testing of the ventilator's performanceon a test lung is planned to ensure compliancewith FDA standards.

6. Acknowledgements

The design and fabrication of this devicewas a term project in MIT Course 2.75:Precision Machine Design. We are grateful toLynn Osborn and Dr. Tom Brady of The Centerfor Integration of Medicine and InnovativeTechnology www.cimit.org for providingsupport for course 2.75 and this project; and theVA Medical Center-West Roxbury. CIMITsupport comes from DOD funds with FAR52.277-11. Special thanks are due to Prof. AlexSlocum, Nevan Hanumara, Conor Walsh andDave Custer for continuous guidance for theproject. We are also thankful to Dr. BarbaraHughey for her help with instrumentation for ourbench level experiments. The authors would also

like to thank SolidWorks Corporation forproviding the solid model software used tocreate the models for this project.

7. References

[1] At-Khaled N, Enarson D, Bousquet J(2001) Chronic respiratory diseases indeveloping countries: the burden andstrategies for prevention and management,Bulletin of the World Health Organization,79 (10)

[2] Chan-Yeung M, Ait-Khaled N, White N, IpMS, Tan WC (2004) The burden andimpact of COPD in Asia & Africa. Int JTuberc Lung Dis.

[3] Webster's New WorldTM MedicalDictionary First, Second and Third Editions(May, 2008) John Wiley & Sons, Inc.

[4] McNeil, DG (2006) Hospitals short ofventilators if bird flu hits, NYT, May 12

[5] Klein KR, Nagel NE (2007) Mass medicalevacuation: Hurricane Katrina and nursingexperiences at the New Orleans airport.Disaster Manag Response. 2007 Apr-Jun;5(2):56-61

[6] deBoisblanc, B.P. (2005) Black Hawk,please come down: reflections on ahospital's struggle to survive in the wake ofHurricane Katrina. Am J Respir Crit CareMed. Nov 15;172(10):1239-40

[7] http://www.news-medical.net/news/20091020/CDC-stockpiles-ventilators-for-use-during-public-health-emergency.aspx

[8] Kerechanin CW, Cytcgusm PN, Vincent JA,Smith DG, Wenstrand DS (2004)Development of Field Portable VentilatorSystems for Domestic and MilitaryEmergency Medical Response, JohnHopkins Apl. Tech. Digest Vol 25, No. 3

[9] ASTM F920-93(1999)