An Energy-Efficient Human-Powered Hydraulic Bicycle With

Article The PurdueTracer: An Energy-Efﬁcient Human-Powered Hydraulic Bicycle with Flexible Operation and Software Aids

Gianluca Marinaro 1,* ID , Zhuangying Xu 2, Zhengpu Chen 3, Chenxi Li 3 ID , Yizhou Mao 2 and Andrea Vacca 2,3,4

1 Department of Industrial Engineering, University Federico II, 80100 Naples, Italy 2 School of Mechanical Engineering, Purdue University, West Lafayette, IN 47906, USA; [email protected] (Z.X.); [email protected] (Y.M.); [email protected] (A.V.) 3 Department of Agricultural and Biological Engineering, Purdue University, West Lafayette, IN 47906, USA; [email protected] (Z.C.), [email protected] (C.L.) 4 Maha Fluid Power Research Center, Purdue University, West Lafayette, IN 47906, USA * Correspondence: [email protected]; Tel.: +39-081-7683511

Received: 8 January 2018; Accepted: 26 January 2018; Published: 31 January 2018

Abstract: Hydrostatic transmissions (HT) are widely applied to heavy-duty mobile applications because of the advantages of layout flexibility, power to weight ratio, and ease of control. Though applications of fluid power in light-duty vehicles face challenges, including the unavailability of off-the-shelf components suitable to the power scale, there are potential advantages for HTs in human-powered vehicles, such as bicycles, the most important one being the energy-saving advantage achievable through regenerative braking in a hybrid HT. This paper describes an innovative design for a hydraulic hybrid bicycle, i.e., the PurdueTracer. The PurdueTracer is an energy-efficient human-powered hydraulic bicycle with flexible operation and software aids. An open-circuit hydraulic hybrid transmission allows PurdueTracer to operate in four modes: Pedaling, Charging, Boost, and Regeneration, to satisfy users’ need for different riding occasions. An aluminum chassis that also functions as a system reservoir was customized for the PurdueTracer to optimize the durability, riding comfort, and space for components. The selection of the hydraulic components was performed by creating a model of the bicycle in AMESim simulation software and conducting a numerical optimization based on the model. The electronic system equipped users with informative feedback showing the bicycle performance, intuitive execution of functions, and comprehensive guidance for operation. This paper describes the design approach and the main results of the PurdueTracer, which also won the 2017 National Fluid Power Association Fluid Power Vehicle Challenge. This championship serves to prove the excellence of this vehicle in terms of effectiveness, efficiency, durability, and novelty.

Keywords: hydrostatic transmission; hydraulic hybrid; light-duty system; energy recovery; software interface; computer-aided monitoring; Internet of Things

1. Introduction Fluid power technology has advantages of layout flexibility, a good power to weight ratio, and ease of control that are well reflected in hydrostatic transmissions and have been successfully applied in many heavy-duty applications including construction, agricultural, forestry and military machines, etc. However, the integration of fluid power technology into light-duty applications still presents significant challenges, especially for human-powered vehicles, such as bicycles. Most of the

Energies 2018, 11, 305; doi:10.3390/en11020305 www.mdpi.com/journal/energies Energies 2018, 11, 305 2 of 24 challenges relate to the unavailability of hydraulic pumps and motors suitable to the power scale and range of operating conditions typical of such applications. Notwithstanding, there are potential advantages to ﬂuid power in human-powered vehicles. In particular, with respect to traditional “pure mechanical” bicycles based on the chain sprocket transmission system, the use of a hybrid hydrostatic transmission allows for energy recovery during braking phases and enables power management strategies, which decouple the human power input from the instantaneous power request given by the vehicle resistance. In particular, the energy stored within the transmission system during phases of braking or of excessive input energy could be utilized in subsequent phases, such as vehicle starts or uphill pedaling conditions, to assist the cyclist. Moreover, a properly designed ﬂuid power system can avoid the typical disadvantages of chain drives, which are:

• The safety risk to the rider in the form of entangling clothing in the chain and the sprocket due to the fact that the chain drive is typically exposed; • The undesirable occurrence in variable-speed bicycles of shifting to the wrong gear and positioning the chain in an intermediate position between sprockets, which leads to surprising and dangerous slipping of the chain and the consequent unbalancing of the cyclist; • The lubrication requirement of the chain, which attracts dust and dirt, causing aging and decay of the efﬁciency of the system and unpleasant effects due to possible contact with clothing.

The energy efficiency of the transmission system as well as the overall weight, particularly due to the abovementioned unavailability of properly designed fluid power components, represent the main design challenges of a hydrostatic transmission for a human-powered vehicle, with respect to a pure mechanical transmission system. Fluid power technology, in comparison with the electric power technology widely adopted in bicycles nowadays, might appear inferior from the point of view of commercialization and technical pros and cons. A deep technical comparison is outside the scope of this paper. However, it is clear that electric power technology takes advantage of the more intensive effort made by the industry in developing components suitable for the size of the application. Another fact to take into consideration is that essentially all electric bicycles do not truly have zero emissions associated with their use, since they require energy charging from the network. Most of the electric bicycles available on the market do not implement braking energy recovery features, which require more expensive electric gearless hub motors. Disposal of electric batteries, which have a finite life span, and use of rare materials in electric components are additional factors that mean the electric vehicle is not a completely environmentally friendly solution for human-powered vehicles. Instead, a hydraulic bicycle can use components that are for the most part made with non-noble materials (iron, steel, or aluminum), and a working fluid that does not require frequent replacement and can be environmentally friendly (such as the bio-oil used in the present study). Moreover, hydraulic hybrid systems can be more efficient than electric hybrid systems [1], and they offer more opportunities for integrating regenerative braking [2]. The potential advantages of hydraulic bicycles have motivated a certain amount of research effort over the last decades. Basic closed-loop hydrostatic transmission systems, essentially using a hydraulic pump connected to the pedals and a hydraulic motor connected to the rear wheel, were proposed in some patents, even recently. Significant is the patent [3] that describes a solution with variable vane units used as pumps and motor. The solution includes a manual mechanism that varies the displacements of the units to maintain constant system pressure (approximately proportional to the rider torque effort). Many other solutions are conceptually similar, but are based on different design architectures of the positive displacement units used as pumps and as motors: for example, Smith [4] uses a different solution with vane type units; instead, the patent [5] proposes a radial piston pump and a gear motor; Chattin [6] uses external gear units as its pump and motor. Piston units are used in the patents [5,7]. Another significant example of specially designed positive displacement machines Energies 2018, 11, 305 3 of 24 was designed by Brackett [8], who proposed a fixed displacement pump and a variable displacement motor. None of these systems includes brake energy recovery features; they are intended to operate as traditional bicycles. In addition, though a functional transmission is created, no electronic device or software aid is integrated to monitor the performance and efficiency. All of the abovementioned patents describe possible technology solutions, but do not report basic sizing considerations or expected/measured performance. Some technical publications address the problem of sizing a basic of hydrostatic transmission for light duty, bicycle-like applications: Chang and Yao [9] describe the case of an open-circuit electro-hydrostatic transmission, although the results reported in the work show a top speed of only 7 km/h for an overall vehicle weight of 135 kg! In Yang and Zhong [10], another design method is reported, although no experimental results are discussed. Some more recent efforts have aimed at exploring energy recuperation in human-powered vehicles. Interesting patents on compact regenerative systems for bicycles were taken out by Amarantos [11] and Swain et al. [12]. More extensive work on regenerative bicycles is presented in the papers written by Lagwankar [13] and Truong et al. [14]. They both describe the sizing of an open-circuit hybrid transmission, in which different modes of operations, including normal pedaling and regenerative braking, are realized by solenoid valves. Unfortunately, both works show only simulation results, but velocities above 32 km/h are predicted. A similar work was also performed at the University of Minnesota [15]: in this work bent axis piston units are used in an open hydrostatic transmission, with an accumulator used for regenerative, manual charge, and boost mode through manual switches. A prototype was also realized, and the performance was good (maximum pedaling speed of about 17.7 km/h), although the weight amount and distribution, as well as the mechanical linkages of the hydraulic units (two chains/sprockets were used) were not optimized. Another study recently completed at the authors’ institution was documented by Foa et al. [16]: in this case, a four-wheel vehicle was realized with an open-circuit hybrid transmission based on external gear pumps. The vehicle demonstrated excellent capabilities for electronic control of the hydraulic system, but the vehicle performance was quite poor. The present paper aims at further contributing to the field of innovative human-powered vehicles, and shows a design procedure for a bicycle that:

(1) Maximizes velocity while pedaling; (2) Minimizes overall weight; (3) Recovers energy while braking; (4) Allows for a power boost (use of energy from the accumulator); (5) Maximizes human comfort and implements all necessary safety features.

Similarly, to the previously mentioned works by Foa et al. [16] and Wang et al. [15], prototypes were realized to participate in a U.S. national competition for human-powered fluid power vehicles open to engineering schools. The vehicle realized within this work, named the “PurdueTracer”, participated in the 2017 Fluid Power Vehicle Challenge (FPVC) organized by the National Fluid Power Association (NFPA), held in Ames, Iowa in April 2017, and won the overall championship. The FPVC fully assesses the main criteria (1)–(5) specified above, through specific evaluation criteria on the vehicle design as well as purposely designed races. In particular, one race is dedicated to endurance (about 2 m); one race to sprint capabilities; and one to the boost capabilities of the vehicles (utilization of the energy stored in the accumulator). More details on the competition can be found online [17]. The following sections describe the methods utilized to design PurdueTracer. It is important to point out that this research put significant focus on the maximization of the performance parameters of the hydraulic system, namely weight, energy efficiency, and amount of energy recovery. However, other important aspects related to the designing of a hydrostatic transmission where the prime mover is a human being were also considered. EnergiesEnergies 2018, 112018, x, FOR11, 305 PEER REVIEW 4 of 24 4 of 24 crankshaft,Over as awell wide as range the ofcrankshaft pedaling speeds pedal (up radius, to 100 have rev/min, effects depending on the on human the individual), performance. The proportionsthe torque typi generatedcally used by in a humancommercial is almost bicycles constant reflect [4,5,10 accurate,18]. Other studies factors performed also affect thein this torque field [18]. Anotherand the resistance level of ofhuman the cyclist.–machine Primarily, interaction the geometrical that affects configuration the performance of the bicycle has of the significant cyclist is related to theimplications visual aids on and the torque monitoring level: the devices angles and available the distances during between the seat, physic handlebar,al exercise. and crankshaft, It is proven for almostas all well intensive as the crankshaft sport activities pedal radius, (gym have cardio effects on equipment, the human performance. jogging, cycling) The proportions that the ability to typically used in commercial bicycles reflect accurate studies performed in this field [18]. monitor theAnother human level status, of human–machine e.g., the heart interaction rate, thatthe affectsinstantaneous the performance power ofthe consumed, cyclist is related etc. as well as vehicleto performance the visual aids data and monitoring(pedaling devicesspeed, availablevehicle velocity) during the can physical improve exercise. the human It is proven ability for to achieve certainalmost performance all intensive goals. sport activitiesAll these (gym human cardio factors equipment, related jogging, to cycling)the structural that the ability design to monitor of the chassis as well asthe the human ease status,of monitoring e.g., the heart and rate, control the instantaneous of the vehicle power were consumed, addressed etc. in as the well proposed as vehicle design. In particular,performance a smartphone data (pedaling app speed,was developed vehicle velocity) for PurdueTracer can improve the humanto enable ability all to the achieve desirable certain features of performance goals. All these human factors related to the structural design of the chassis as well as the a modernease of and monitoring intelligent and control vehicle, of the which vehicle co wereuld addressed potentially in the encourage proposed design.potential In particular, users to use smart hydraulica smartphone bikes like app the was one developed proposed for in PurdueTracer the research to done enable by all Navarro the desirable et al. features [19]. of a modern and intelligent vehicle, which could potentially encourage potential users to use smart hydraulic bikes 2. Systemlike the Overview one proposed in the research done by Navarro et al. [19]. This2. System section Overview provides a general overview of the system on the basis of the PurdueTracer vehicle developedThis within section this provides study. a general PurdueTracer overview of reproduces the system on the the basis configuration of the PurdueTracer of a traditional vehicle bicycle (Figuredeveloped 1); however, within thisit integrates study. PurdueTracer an electro reproduces-hydraulic the and configuration a hybrid of system a traditional used bicycle for the motion transmission.(Figure1); The however, solenoid it integrates valves determine an electro-hydraulic the configuration and a hybrid of system the hydraulic used for thesystem, motion and they are controlledtransmission. using The an solenoid Arduino valves controller determine that the configuration communicates of the hydraulic via Bluetooth system, and with they a are smartphone controlled using an Arduino controller that communicates via Bluetooth with a smartphone application. application.The overall The electronic overall controlelectronic system control applies system the concept applies of the the Internet concept of Things of the (IoT), Internet providing of theThings (IoT), providingcyclist the with cyclist a smartphone with a applicationsmartphone to control application PurdueTracer, to control with functionalPurdueTracer, aids in promotingwith functional the aids in promotinguser experience. the user experience.

FigureFigure 1.1. PhysicalPhysical design design for PurdueTracer.for PurdueTracer.

The wholeThe whole system system circuit circuit is is represented represented in in Figure Figure2, which 2, which integrates integrates the ISO schematicthe ISO schematic of the of the hydraulichydraulic system, system, the the mechanical mechanical connectionsconnections through through gearboxes gearboxes to the pedals to the and pedals the wheels, and and the the wheels, and electronic controller. All these subsections of the system will be detailed in the following sections of the electronicthe paper. controller. All these subsections of the system will be detailed in the following sections of the paper.

Energies 2018, 11, 305 5 of 24 Energies 2018, 11, x FOR PEER REVIEW 5 of 24

SMARTPHONE DAQ AND CONTROL RELAYS

BRAKE CONTROL BATTERY

BATTERY

FREE WHEEL PEDALING AND DOG GEAR SYSTEM

Low power electric signals High power electric signals Sensors reading

Figure 2. Circuit design of PurdueTracer. Figure 2. Circuit design of PurdueTracer. 3. Hydraulic System Design 3. Hydraulic System Design 3.1. Hydraulic Hybrid Circuits 3.1. Hydraulic Hybrid Circuits The hydraulic schematic of PurdueTracer is shown in Figure3. The circuit is essentially designed basedThe on hydraulic an open-circuit schematic configuration, of PurdueTracer where is a shown fixed displacement in Figure 3. The pump circuit is connectedis essentially to thedesigned pedal basedand provides on an open flow-circuit to a fixed configuration, displacement where motor a fixed connected displacement to the rear pump wheel is throughconnected a variable to the pedal shift andgear provides hub. The flow system to a also fixed includes displacement an accumulator, motor connected which can to storethe rear energy wheel from through the supply a variable (either shift the gearpump hub. connected The system to the also pedal includes or an auxiliaryan accumulator, manual which pump), can or store while energy braking. from the supply (either the pumpThe configuration connected to ofthe the pedal hydraulic or an circuitauxiliary is determinedmanual pump), by two or solenoidwhile braking. switching valves, so that four differentThe configuration modes of operationof the hydraulic can be realized:circuit is Pedaling,determined Charging, by two Boost, solenoid and switching Regeneration. valves, These so thatmodes four are different illustrated modes in Figure of operation3. can be realized: Pedaling, Charging, Boost, and Regeneration. TheseFor modes this are light-duty illustrated application, in Figure 3. the choice of an open-circuit HT, instead of a closed-circuit configuration, which is more common for heavy-duty HTs, has the following merits: V1 V2 A V1 V2 A • It reducesNot energized theNot number energized of components (i.e., charge pumpEnergized circuit).Energized Closed-circuit HTs generally reduce the size of the reservoir. However, for this application, the reservoir size requirement 3W 3W

given by the accumulator chargeA2 and discharge processes turned out to be more stringentA2 than that related to the flow of the smallV2 pump used in the circuit. V2 V1 V1 • It allows easyCV1 implementationA1 of differentCV2 modes of operationCV1 throughA1 switchingCV2 valves, HP HP without significantP layout complications. P A3 M RP A3 M RP Fixed displacement hydrostaticRV unitsDG for both the pump and the motor wereRV preferredDG to variable displacement units for the following reasons: • Limited availability of lightweight material (i.e., aluminum) pumps and motors in the displacement (a) (b) range of interest for this application (<10 cm3/rev, as will be detailed in the Section 3.2). • The use of an electrically controlled multiple-speed gear hub connected to the rear wheel can permit regulation similar to a secondary controlled HT, but avoids the use of a variable unit at partial displacement, where the energy efficiency can be low. • An electro-hydraulic system for adjusting the instantaneous displacement of the units would add increased weight and complexity to the system.

Energies 2018, 11, x FOR PEER REVIEW 5 of 24

SMARTPHONE DAQ AND CONTROL RELAYS

BRAKE CONTROL BATTERY

BATTERY

FREE WHEEL PEDALING AND DOG GEAR SYSTEM

Low power electric signals High power electric signals Sensors reading

Figure 2. Circuit design of PurdueTracer.

3. Hydraulic System Design

3.1. Hydraulic Hybrid Circuits The hydraulic schematic of PurdueTracer is shown in Figure 3. The circuit is essentially designed based on an open-circuit configuration, where a fixed displacement pump is connected to the pedal and provides flow to a fixed displacement motor connected to the rear wheel through a variable shift gear hub. The system also includes an accumulator, which can store energy from the supply (either the pump connected to the pedal or an auxiliary manual pump), or while braking. The configuration of the hydraulic circuit is determined by two solenoid switching valves, so Energiesthat2018 four, 11 different, 305 modes of operation can be realized: Pedaling, Charging, Boost, and Regeneration.6 of 24 These modes are illustrated in Figure 3.

V1 V2 A V1 V2 A Not energized Not energized Energized Energized

3W 3W

A2 A2

V2 V2 V1 V1 CV1 A1 CV2 CV1 A1 CV2 HP HP P P

A3 M RP A3 M RP RV DG RV DG

Energies 2018, 11, x FOR PEER REVIEW 6 of 24 (a) (b)

V1 V2 A V1 V2 A Not energized Energized Not energized Energized

3W 3W

A2 A2

V2 V2 V1 V1 CV1 A1 CV2 CV1 A1 CV2 HP HP P P A3 M RP A3 M RP RV DG RV DG

Figure 3. Hydraulic circuit and valve conditions for (a) Pedaling Mode; (b) Charging Mode; (c) Boost Figure 3. Hydraulic circuit and valve conditions for (a) Pedaling Mode; (b) Charging Mode; (c) Boost Mode; and (d) Regeneration Mode. Note: Check Valve 1 (CV1); Check Valve 2 (CV2); Relieve Valve Mode; and (d) Regeneration Mode. Note: Check Valve 1 (CV1); Check Valve 2 (CV2); Relieve Valve (RV); (RV); 3-Way L Type Valve (3W); Solenoid Valve 1 (V1), normally open; Solenoid Valve 2 (V2), 3-Waynormally L Type Valve closed; (3W); Accumulator Solenoid Valve(A); Dog 1 (V1), Gear normally (DG); Hand open; Pump Solenoid (HP); Valve Pump 2 (V2), (P); normally Motor (M); closed; AccumulatorRegeneration (A); Pump Dog Gear (RP). (DG); Hand Pump (HP); Pump (P); Motor (M); Regeneration Pump (RP).

EachFor operating this light mode-duty of application, PurdueTracer the illustrated choice of an in open Figure-circuit3 deserves HT, instead a more of detailed a closed description,-circuit configuration, which is more common for heavy-duty HTs, has the following merits: as follows: Pedaling It reduce(Figures the3 a).number When of neithercomponents of the (i.e. solenoid, charge pump valves circuit). V1 and Closed V2 is-circuit energized, HTs generally the vehicle operates inreduce this mode,the size replicating of the reservoir. the functioning However, for of athis traditional application, bicycle. the reservoir This implies size requirement that the vehicle can also begiven operated by the accumulator in this mode charge in caseand discharge of insufficient processes battery, turned or out in to the be casemore ofstringent a failure than of the that related to the flow of the small pump used in the circuit. electronic control system. In Pedaling Mode, the rider spins the pump shaft through the pedals,  It allows easy implementation of different modes of operation through switching valves, and the pump flow is sent directly to the motor connected to the rear wheel. without significant layout complications. Charging (Figure3b). This mode is realized by switching both V1 and V2. In this case, two options are availableFixed for displacement charging. Hand hydrostatic pump charging: units for both in this the case, pump the and vehicle the motor is at rest were and preferred the cyclist to uses variable displacement units for the following reasons: his/her legs to balance the bicycle and hands to charge the accumulator. A check valve prevents the fluid under Limited pressure availability to flow of through lightweight the pumpmaterial and (i.e. spin, aluminum) the crankshaft pumps pedals. and motors The hand in the pump chargingdisplacement mode is used range in of this interest project, for this mainly application because (<10 itcm was3/rev, used as will during be detailed the FPVC.in the Section However, another charging3.2). mode is available with the same configuration for V1 and V2, i.e., Pedaling charge:  The use of an electrically controlled multiple-speed gear hub connected to the rear wheel can in this case, the cyclist pedals to pump flow to the accumulator. This mode can be useful for storing permit regulation similar to a secondary controlled HT, but avoids the use of a variable unit at energy in passive riding phases, such as riding downhill. partial displacement, where the energy efficiency can be low. Boost An(Figure electro3c).-hydraulic The boost system mode for is adjusting realized whenthe instantaneous only V2 is energized.displacement This of isthe the units mode would used to release theadd energy increased stored weight in the and accumulator. complexity to Thethe system. boost mode can be used when the cyclist wants to move the bike without pedaling. RegenerationEach operating(Figure mode3d). An of additional PurdueTracer pump illustrated driven in by Figure the rear 3 wheel deserves can a be more engaged detailed by the riderdescription through a, dog as follow gears mechanism: that will be detailed in Section4. The pump generates flow to the accumulator,Pedaling and (Figure the shaft 3a). torque, When neither which of corresponds the solenoid to valves the braking V1 and torqueV2 is energized, at the wheel, the vehicle is defined operates in this mode, replicating the functioning of a traditional bicycle. This implies that the vehicle by the pressure level in the accumulator. A pressure relief valve, with adjustable pressure setting can also be operated in this mode in case of insufficient battery, or in the case of a failure of the through a proportional solenoid, is used to limit the maximum braking torque at the level selected by electronic control system. In Pedaling Mode, the rider spins the pump shaft through the pedals, and the cyclist.the pump As flow will is be sent detailed directly in to Section the motor4, inconnected this work to the the rear regeneration wheel. system was sized in such Charging (Figure 3b). This mode is realized by switching both V1 and V2. In this case, two options are available for charging. Hand pump charging: in this case, the vehicle is at rest and the cyclist uses his/her legs to balance the bicycle and hands to charge the accumulator. A check valve prevents the fluid under pressure to flow through the pump and spin the crankshaft pedals. The hand pump charging mode is used in this project, mainly because it was used during the FPVC. However, another charging mode is available with the same configuration for V1 and V2, i.e., Pedaling charge: in this

Energies 2018, 11, 305 7 of 24 a way that the whole energy of a speciﬁc braking event can be stored in the accumulator. However, in case of more demanding braking events, or prolonged braking (downhill phases), the regeneration circuit becomes purely dissipative, since the pump ﬂow is relieved through the pressure relief valve. The use of an additional pump for the regeneration circuit could be avoided by using a single pump/motor unit connected to the rear wheel. However, the solution based on the additional pump was preferred because:

• It permitted calculating the motor size independently from the needs of the regeneration circuit. In fact, the size of the hydraulic motor was determined based on the requirements of the Pedaling and Boost modes. A less advantageous size would have resulted if the needs of the Regeneration mode’s function were also considered. • It permitted us to use a “free wheel” gear hub, like in a traditional bike. A free wheel hub allows for unidirectional transmission of the torque transmitted to the wheels, and allows for vehicle coasting (wheel spinning without involving angular velocity at the gear hub). A single pump/motor solution would have required a bidirectional hub, with the undesired consequence of eliminating the coasting function. In other words, the cyclist would feel the parasitic loss of the hydraulic motor while coasting.

While the proportional RV has a functional role in Regeneration mode, it also has a safety role in the remaining modes (Pedaling, Charge, and Boost). For this reason, the input signal to the RV solenoid is chosen in such a way that it reduces the RV setting pressure. An additional safety element is given by the valve 3W, which is used to manually relieve the accumulator pressure. 3W is always closed during the normal operation of the vehicle.

3.2. Simulation and Sizing of the System The hydraulic layout described in the previous section offers all the functions suitable for a human-powered vehicle. However, good operation in all four modes can only be realized through proper sizing of the key components of the hydraulic circuit. In particular, the size of the positive displacement units, as well as the gearboxes at both the pedals and at the rear wheel, needs to be found to guarantee not only maximum vehicle velocity while pedaling, but also an effective boost and regeneration function. The optimal sizing of PurdueTracer was obtained through a numerical optimization. The following sub-sections ﬁrst detail the approach utilized for the numerical modeling of the vehicle, and then describe the formulation of the optimization problem, considering the typical performance of a cyclist.

3.2.1. Numerical Model of the PurdueTracer The numerical model of the PurdueTracer was realized through the multidisciplinary software AMESim by Siemens. The model utilizes a quasi-steady state approach, and uses these well-known relations for the hydrostatic units: Qp = ηv,p·np·VD,p (1)

nm·VD,m Qm = (2) ηv,m

VD,p·∆pp Tp,e = (3) ηhm,p

Tm,e = VD,m·∆pm·ηhm,m (4) For the accumulator, the polytrophic relation for nitrogen gas is used:

n pacc·Vgas,acc = constant. (5) Energies 2018, 11, x FOR PEER REVIEW 8 of 24

푛푚 ∙ 푉퐷,푚 푄푚 = (2) 휂푣,푚

푉퐷,푝 ∙ ∆푝푝 푇푝,푒 = (3) 휂ℎ푚,푝

푇푚,푒 = 푉퐷,푚 ∙ ∆푝푚 ∙ 휂ℎ푚,푚. (4) For the accumulator, the polytrophic relation for nitrogen gas is used: Energies 2018, 11, 305 8 of 24 푛 푝푎푐푐 ∙ 푉푔푎푠,푎푐푐 = 푐표푛푠푡푎푛푡. (5)

TheThe value of the polytrophic constant n was assumed to be ( n = 1.2), 1.2), representative representative of of a a process process with a certain amount of heat transfer for the gas. The The volume volume of of the the accumulator accumulator is is related related to to the the prepre-charge-charge pressure: 1 p푝max 푛n ( 푚푎푥) ∙·Δ∆푉V p푝 V 푉= = min푚𝑖푛 acc 푎푐푐 1 1 11 . .(6 (6)) 푝 n 푛 푝 푛n p(o 표 ) ∙[( p푚푎푥max ) −1] 푝 · 푝 − 1 pmin푚𝑖푛 p푚𝑖푛min

AsAs is shown in the AMESimAMESim modelmodel schematicschematic provided provided in in Figure Figure4, 4, the the main main pump pump is connectedis connected to tothe the pedals pedals with with a gearbox a gearbox of variableof variable transmission transmission ratio, ratio, while while the motorthe motor is connected is connected to the to wheel the wheel with witha gearbox a gearbox of variable of variable gear ratio. gear Bothratio. gearboxes Both gearboxes are assumed are assumed to be ideal to (unitbe ideal efﬁciency). (unit efficiency). The hydraulic The hydraulicvalves are valves assumed are ideal assumed (no pressure ideal (no losses), pressure as well losses), as the as connecting well as the lines. connecting Realistic volumeslines. Realistic for the volumesconnecting for lines the connecting are assumed lines to considerare assumed their to capacitance consider their effects. capacitance effects.

Figure 4. AMESim model schematic for the entire bicycle. Figure 4. AMESim model schematic for the entire bicycle.

To make the model suitable to perform design considerations and assist in the selection of the size forTo both make the the hydraulic model suitable pump toand perform motor, designrealistic considerations efficiency maps and were assist used in as the shown selection in Figure of the 5.size These for bothmaps the came hydraulic from experiments pump and motor, performed realistic on efﬁciencya small gear maps unit were (of about used as4 cm shown3/rev). in Figure5. These maps came from experiments performed on a small gear unit (of about 4 cm3/rev). The model permits us to reproduce all modes of operation displayed in Figure3. For the Pedaling mode, the crankshaft speed serves as the input parameter, and the simulation returns the vehicle velocity through a road friction model. The acceleration of the vehicle avehi is calculated by the following equation:

1 a = . F − Faero − F − F . (7) vehi Iw W slope f riction mvehi + 2 rW

The propelling force FW applied on the rear wheel with the external radius rW and the moment of inertia IW is: TW FW = . (8) rW Energies 2018, 11, 305 9 of 24

The aerodynamic force Faero is given by the equation:

1 F = ρ S C (V + V )2. (9) aero 2 air x x vehi wind

The slope force Fslope follows Equation (10):

Fslope = (mvehi + mcyclist)g sin α. (10)

Finally, the friction force Ff riction is given by the following equation:

Ff riction = f Vvehi + µr(mvehi + mcyclist)g cos α. (11) Energies 2018, 11, x FOR PEER REVIEW 9 of 24

(a)

(b)

Figure 55.. EfficiencEfficiencyy plots plots for for hydraulic hydraulic pump pump and and motor: motor: (a) hydro(a) hydro-mechanical-mechanical efficiency; efficiency; (b) volumetric(b) volumetric efficiency efficiency..

The model permits us to reproduce all modes of operation displayed in Figure 3. For the 3.2.2. Numerical Optimization Pedaling mode, the crankshaft speed serves as the input parameter, and the simulation returns the vehicleThe velocity model through described a road in thefriction previous model. section was used to perform the sizing of the main componentsThe acceleration of the system, of the vehicle by means 푎푣푒ℎ𝑖 ofis acalculate numericald by optimizationthe following method.equation: The inputs of the optimization problem are: the unit displacements1 of the pump and motor, VD,p, VD,m, the gearboxes 푎 = . (퐹 − 퐹 − 퐹 − 퐹 ) 푣푒ℎ𝑖 퐼푤 푊 푎푒푟표 푠푙표푝푒 푓푟𝑖푐푡𝑖표푛 . (7) ratios of the pump and motor, rp, 푚rm푣푒,ℎ and𝑖+ the2 pre-charge pressure of the accumulator, p0. 푟푊 Assumptions used for the simulation are: the human speed of pedaling (70 rpm [20]), the mass of The propelling force 퐹 applied on the rear wheel with the external radius 푟 and the moment the vehicle (mvehi = 140 kg; this푊 value considers the estimated mass of the hydraulic푊 system, the frame, ofand inertia the cyclist), IW is: the slope (α = 1◦, assumed to meet the maximum slope of the track used during the ﬁnal FPVC competition), the accumulator size (Vacc =푇 2푊 L, which is the size of a low-weight carbon-ﬁber 퐹푊 = . (8) 푟푊

The aerodynamic force 퐹푎푒푟표 is given by the equation: 1 퐹 = 휌 푆 퐶 (푉 + 푉 )2. 푎푒푟표 2 푎𝑖푟 푥 푥 푣푒ℎ𝑖 푤𝑖푛푑 (9)

The slope force 퐹푠푙표푝푒 follows Equation (10):

퐹푠푙표푝푒 = (푚푣푒ℎ𝑖 + 푚푐푦푐푙𝑖푠푡)푔 sin 훼. (10)

Finally, the friction force 퐹푓푟𝑖푐푡𝑖표푛 is given by the following equation:

퐹푓푟𝑖푐푡𝑖표푛 = 푓푉푣푒ℎ𝑖 + 휇푟(푚푣푒ℎ𝑖 + 푚푐푦푐푙𝑖푠푡)푔 cos 훼. (11)

3.2.2. Numerical Optimization The model described in the previous section was used to perform the sizing of the main

Energies 2018, 11, 305 10 of 24 accumulator that was made available to the authors for the ﬁnal implementation), the maximum accumulator pressure, pmax, resulting from the charging model (pmax = 180 bar, from the hand pump 2 charging process), the wheel radius and moment of inertia, (rw = 0.334 m, Iw = 0.15 kg·m ), the initial velocity of the vehicle in the Regeneration mode (vo,reg = 12.8 km/h), and the initial accumulator pressure (po = 25 bar). Optimization constraints were represented by the maximum torque at the pedals, limited to 25 N·m (a reasonable maximum torque that can be input by a human [20]), so that all the designs that exceed this maximum torque value will be rejected by the optimization algorithm. The objective functions (OF) of the optimization problem are deﬁned as follows: OF1: max vpedal

OF2: max (boost).

The parameter boost is deﬁned as:

m × L boost = vehi . (12) p0,race × Vacc

The objective function for the regeneration is deﬁned as follows:

OF3: max (pacc).

OF1 has a straightforward interpretation: it simply tends to maximize the bike velocity during the pedaling mode. Instead with OF2, the optimization seeks for a solution that gives the maximum effectiveness of the accumulator during boost mode. The definition of OF2 is also in accordance with the rules defined by the FPVC competition [21]. OF3 was instead used to reach the maximum energy storage during regeneration mode. To perform this optimization, the AMESim model of the vehicle in Figure4 was run to reproduce the pedaling model and the boost mode (discharging the accumulator from the pmax level) for every design, given by a combination of input parameters. The non-linear programming by quadratic Lagrangian (NLPQL) optimization algorithm, present in the AMESim software (version 15), was utilized for this purpose. In particular, a first genetic algorithm optimization is performed to explore a wide range of input parameters. Afterwards, local NLPQL sub-optimizations were performed to better explore the local minima, properly refining the range of the input parameters, to get more accurate results.

3.2.3. Optimization Results and Predicted Performance The best design resulting from the optimization process is shown in Table1. The same table reports the actual values chosen for the PurdueTracer, which are the closest values of practical components available to the authors.

Table 1. Optimization results and component selection.

Best Design Selected Components Component Size Company Size Pump displacement 4.52 cc/rev CASAPPA 4.27 cc/rev Motor displacement 2.13 cc/rev CASAPPA 2.13 cc/rev Accumulator volume 2.00 L STEELHEAD COMPOSITES 2.00 L Front gear ratio 1/5.68 MISUMI 1/6.32 Rear gear ratio 4.00 MISUMI 4.00 Regeneration gear ratio 17.82 ANDYMARK 16.80 Regeneration pump displacement 4.23 cc/rev CASAPPA 4.27 cc/rev Accumulator pre-charge gas pressure 25 bar - 25 bar Energies 2018, 11, 305 11 of 24

Good performance in terms of maximum velocity during pedaling and boost effectiveness was reached. As pertains to the increase of accumulator pressure during regeneration (OF3), the value is limited due to the low momentum of the vehicle during a single full-stop braking event. Higher potentials for energy recovery could be achieved considering a longer braking phase (such as controlling speed while traveling downhill). The predicted performance of the vehicle with the selected components is summarized in Table2.

Table 2. Results of best design acquired from optimization and selected components.

Performances Performances in Pedaling Mode Best Design from Optimization Selected Components Power 183 W 223 W Torque IN (Human) 25 N·m 30 N·m Pump shaft 435 rpm 442 rpm Bike speed 5.10 m/s 5.87 m/s Main line pressure 45.93 bar 64.59 bar Mainline flow rate 1.81 L/min 1.64 L/min Pump volumetric efficiency 88.91% 86.36% Pump mechanical efficiency 86.76% 90.85% Motor volumetric efficiency 94.62% 90.81% Motor mechanical efficiency 85.55% 90.43% Hydraulic transmission efficiency 62.44% 64.44% Performances in Boost Mode Best Design from Optimization Selected Components Max speed 5.21 m/s 4.86 m/s Efficiency function (OF2) 51.1232 50.5471 Distance covered 221 m 214 m

Table2 reports the values with the highest gear ratio of the Shimano gear hub at the rear gear, which actually provides a torque request that exceeds the 25 N·m limit by 5 N·m. Therefore, the variable gear hub at the rear wheel (this will be further explained in Section4) permits a certain ﬂexibility in the strength of the cyclist. If the cyclist can input a lower torque, the parameters such as power and speed (Table2) will be reduced accordingly. The simulation model is suitable to perform transient analyses of the vehicle operation. For example, Figure6 reports the case of a simulation where the accumulator is ﬁrst charged to 200 bars for 100 s in Charging Mode. After this initial phase, a Boost Mode is simulated to replicate the vehicle acceleration. Then, the cyclist starts to pedal the vehicle at 220 s and reaches a constant speed at 280Energies s. Finally, 2018, 11 the, x FOR regeneration PEER REVIEW system is activated at 320 s, stopping the vehicle within 1012 of s. 24

Accumulator Pressure (bar) Vehicle Velocity (m/s)

250.00 7 6 200.00 5 150.00 4

100.00 3 Velocity (m/s) Velocity Pressure (bar) Pressure 2 50.00 1 0.00 0 0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 Time (s)

Figure 6. Results of optimization and simulation models. Figure 6. Results of optimization and simulation models. 4. Mechanical System Design

4.1. Frame The mechanical configuration of the PurdueTracer (Figure 1) was designed to achieve safety, durability, comfortable riding, and optimal weight. The frame accommodates all the hydraulic components listed in Figure 3 and the elements necessary for the electronic control and DAQ system. The lengths and the angular proportion of the bike were chosen with respect to those of the traditional bike shown in Figure 7a. In particular, the seat angle measured as the angle between the seat tube and the horizontal ground was 74°. According to Burke [20], this increases pedaling comfort for the cyclist, who can achieve optimal posture and control of the bike, and maximizes the torque output by the pedals.

(a) (b)

Figure 7. Bike frame diagram: (a) Traditional bike frame; (b) bike frame of PurdueTracer.

As is visible in Figure 1, all hydraulic components are located close to the central frame bar, near the lowest region of the structure. This allows us to lower the center of gravity of the bicycle. An important element of any hydraulic transmission system is the reservoir. In this case the reservoir is integrated within the frame of the bicycle. In particular, the central frame (seat tube) and the down tube of the bike are hollow beans that function as a system reservoir (Figure 7b). The overall volume of the reservoir is Vres = 3.7 L. Such volume allows the working fluid to remain within the reservoir for a sufficient time during pedaling (t > 1 min). This is achieved by a proper configuration of the return lines (at the top of the frame) and the pump intake line (at the bottom of the frame). In this way, incondensable gases that accumulate in the working fluid can be released within the tank, through a vent installed at the top of the down tube. The size of the reservoir also allows for complete

Energies 2018, 11, x FOR PEER REVIEW 12 of 24

Accumulator Pressure (bar) Vehicle Velocity (m/s)

250.00 7 6 200.00 5 150.00 4

100.00 3 Velocity (m/s) Velocity Pressure (bar) Pressure 2 50.00 1 0.00 0 0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 Time (s)

Energies 2018, 11, 305 12 of 24 Figure 6. Results of optimization and simulation models.

4.4. Mechanical Mechanical System System Design

4.1.4.1. Frame Frame TheThe mechanical mechanical configuration conﬁguration of the PurdueTracer (Figure(Figure1 )1) was was designed designed to to achieve achieve safety, safety, durability,durability, comfortable riding,riding, andand optimal optimal weight. weight. The frame frame accommodates accommodates all all the the hydraulic hydraulic componentscomponents listed listed in in Figure Figure 33 andand thethe elementselements necessarynecessary for for the the electronic electronic control control and and DAQ DAQ system. system. TheThe lengths lengths and and the angular proportion ofof thethe bikebike werewere chosenchosen withwith respectrespect to to those those of of the the traditionaltraditional bike bike shown shown in in Figure 77a.a. InIn particular,particular, the the seat seat angle angle measured measured as as the the angle angle between between the the ◦ seatseat tube tube and and the the horizontal horizontal ground waswas 7474°.. According toto BurkeBurke [[20],20], thisthisincreases increases pedaling pedaling comfort comfort forfor the the cyclist, cyclist, who can achieve optimal postureposture andand controlcontrol ofof thethe bike,bike, and and maximizes maximizes the the torque torque outputoutput by by the the pedals. pedals.

(a) (b)

FigureFigure 7 7.. BikeBike frame frame diagram: diagram: ( (aa)) Traditional Traditional bikebike frame;frame; ((bb)) bikebike frame frame of of PurdueTracer. PurdueTracer.

As is visible in Figure 1, all hydraulic components are located close to the central frame bar, near As is visible in Figure1, all hydraulic components are located close to the central frame bar, the lowest region of the structure. This allows us to lower the center of gravity of the bicycle. An near the lowest region of the structure. This allows us to lower the center of gravity of the bicycle. important element of any hydraulic transmission system is the reservoir. In this case the reservoir is An important element of any hydraulic transmission system is the reservoir. In this case the reservoir integrated within the frame of the bicycle. In particular, the central frame (seat tube) and the down is integrated within the frame of the bicycle. In particular, the central frame (seat tube) and the tube of the bike are hollow beans that function as a system reservoir (Figure 7b). The overall volume down tube of the bike are hollow beans that function as a system reservoir (Figure7b). The overall of the reservoir is Vres = 3.7 L. Such volume allows the working fluid to remain within the reservoir volume of the reservoir is Vres = 3.7 L. Such volume allows the working fluid to remain within the for a sufficient time during pedaling (t > 1 min). This is achieved by a proper configuration of the reservoir for a sufficient time during pedaling (t > 1 min). This is achieved by a proper configuration return lines (at the top of the frame) and the pump intake line (at the bottom of the frame). In this of the return lines (at the top of the frame) and the pump intake line (at the bottom of the frame). way, incondensable gases that accumulate in the working fluid can be released within the tank, In this way, incondensable gases that accumulate in the working fluid can be released within the tank, through a vent installed at the top of the down tube. The size of the reservoir also allows for complete through a vent installed at the top of the down tube. The size of the reservoir also allows for complete charging/discharging phases of the accumulator during charging and boost modes. This design of the reservoir not only eliminates the use of an external reservoir, but also allows for optimal weight distribution, limiting oscillations of the center of gravity due to fluid flushing. The material used for the frame is aluminum, and the overall weight of the frame is 9.2 kg. The structural resistance of the frame was verified through FEM analysis, as shown in Figure8. The constraints applied to the model are the central node of the front fork, allowing only displacements in the direction of motion, and the central node wheel fork, allowing no displacement. Based on the maximum strength (11.67 MPa) employed on the frame, the material of the frame was selected as aluminum 6061, which has a yield strength of 55 MPa. Maestrelli and Falsini studied the load applied to a bike [21], using it to evaluate different behaviors of the frame during the pedaling. Static analyses have been performed, applying four types of loads. Energies 2018, 11, x FOR PEER REVIEW 13 of 24 charging/discharging phases of the accumulator during charging and boost modes. This design of the reservoir not only eliminates the use of an external reservoir, but also allows for optimal weight distribution, limiting oscillations of the center of gravity due to fluid flushing. The material used for the frame is aluminum, and the overall weight of the frame is 9.2 kg. The structural resistance of the frame was verified through FEM analysis, as shown in Figure 8. The constraints applied to the model are the central node of the front fork, allowing only displacements in the direction of motion, and the central node wheel fork, allowing no displacement. Based on the Energiesmaximum2018 ,strength11, 305 (11.67 MPa) employed on the frame, the material of the frame was selected13 of as 24 aluminum 6061, which has a yield strength of 55 MPa.

Figure 8. Sample of 0FEA Analysis for the Frame of PurdueTracer, in case of sitting, pedaling, and Figure 8. Sample of 0FEA Analysis for the Frame of PurdueTracer, in case of sitting, pedaling, pushing on the right pedal loads. and pushing on the right pedal loads.

Maestrelli and Falsini studied the load applied to a bike [21], using it to evaluate different behaviorsThe load of the cases frame analyzed during arethe thepedaling. following: Static analyses have been performed, applying four types of• loads.Sitting pedaling, pushing on the right/left pedal: in which the rider is sitting on the saddle and pushingThe load oncases the analyzed right/left are pedal the infollowing: an analysis (depending on the case analyzed); • Braking:Sitting pedaling, in which pushing the rider on is the braking right/left with pedal: both brakes; in which the rider is sitting on the saddle and • Roadpushing irregularity: on the right/left in which pedal the in loads an analysis simulate (depending the road deformity. on the case analyzed);  Braking: in which the rider is braking with both brakes; From the analysis results it appears that most of the structures are under small stress (about 0–3 MPa),  Road irregularity: in which the loads simulate the road deformity. indicating the whole frame is stable. Looking at places with relatively high stress, a factor of safety is calculated.From the analysis The maximum results it stress appears (16.64 that MPa)most of is the present structures during are the under road small instability stress (about condition. 0–3 ThisMPa), condition indicating gives the whole a factor frame of safety is stable. of 3.31, Looking which at is places considered with torelatively be enough high for stress, the considereda factor of prototypesafety is calculated. to function The under maximum all possible stress operating (16.64 MPa) conditions. is present during the road instability condition. This Thecondition layout gives of the a componentsfactor of safety in theof 3.31, frame which is visible is considered from Figure to 1be: the enough main pumpfor the isconsidered mounted prototypenext to the to pedals, function while under the all hand possible pump operating is installed conditions. at the top tube of the frame. The accumulator is placedThe behind layout the of seatthe components tube. The regeneration in the frame pump is visible and hydraulicfrom Figure motor 1: the are main located pump next is to mounted the rear wheelnext to hub. the pedals, while the hand pump is installed at the top tube of the frame. The accumulator is placed behind the seat tube. The regeneration pump and hydraulic motor are located next to the rear wheel4.2. Gearboxes hub. The gearboxes of PurdueTracer implement the gear ratios found by the numerical optimization

(Table1). All the gear boxes were implemented using external gears, in assemblies that ensure compactness and no axial loads at the gear shafts. Figure9 illustrates the mechanical designs of the gearbox. The front gear box ratio is supported by the lower part of the seat tube and approximates the numerical results with a 19/120 gear—integrals gear teeth—thus realizing a ratio of 1:6.3 (Figure9a). The rear gearbox connected to the hydraulic motor is located at on side of the rear wheel, and it is implemented by a 100/17 gear pair, thus implementing a 5.9:1 ratio, close to the numerical optimization. Energies 2018, 11, x FOR PEER REVIEW 14 of 24

4.2. Gearboxes Energies 2018, 11, 305 14 of 24 The gearboxes of PurdueTracer implement the gear ratios found by the numerical optimization (Table 1). All the gear boxes were implemented using external gears, in assemblies that ensure Ancompactness aluminum and plate no is axial used loads to rigidly at the constrain gear shafts. the axisFigure of the 9 illustrates hydraulic the motor mechanical with the hubdesigns of the of rearthe wheelgearbox. (Figure 9b).

Figure 9. 9. GearboxGearbox designs: designs: (a ) (a Front) Front gearbox; gearbox (b); Motor(b) Motor gearbox gearbox and (c , andd) Regeneration (c,d) Regeneration system gearbox. system gearbox. Besides the motor-to-wheel transmission, an electronic 8-Speed gear hub (Shimano Alﬁne) was The front gear box ratio is supported by the lower part of the seat tube and approximates the installed to permit a variable gear ratio. This internal gear hub provides eight different gear ratios numerical results with a 19/120 gear—integrals gear teeth—thus realizing a ratio of 1:6.3 (Figure 9a). from a minimum of 0.5:1 to a maximum of 1.6:1. The variable gear ratio was chosen to permit a wider The rear gearbox connected to the hydraulic motor is located at on side of the rear wheel, and it is range of utilization of the vehicle, around the optimal point found by the optimization. This feature implemented by a 100/17 gear pair, thus implementing a 5.9:1 ratio, close to the numerical allows us to adapt the hydraulic transmission system to different cyclists or different road slopes. optimization. An aluminum plate is used to rigidly constrain the axis of the hydraulic motor with the The gearbox for the regeneration pump (Figure9c,d) implements the optimal ratio of 16.8:1, hub of the rear wheel (Figure 9b). found numerically by means of a two-stage transmission (Table3). Figure9c shows the engagement Besides the motor-to-wheel transmission, an electronic 8-Speed gear hub (Shimano Alfine) was of the dog gear and dog. In this case, the regeneration system is functioning. In normal operation, installed to permit a variable gear ratio. This internal gear hub provides eight different gear ratios the dog gear and dog are separate from each other. The dog is controlled by a hand level that pushes from a minimum of 0.5:1 to a maximum of 1.6:1. The variable gear ratio was chosen to permit a wider the dog back and forth along the shaft. This simple system allows us to avoid parasitic losses of the range of utilization of the vehicle, around the optimal point found by the optimization. This feature regeneration pump, since this component is turned on only when required by the cyclist. allows us to adapt the hydraulic transmission system to different cyclists or different road slopes. The gearbox for the regeneration pump (Figure 9c,d) implements the optimal ratio of 16.8:1, Table 3. Regeneration pump gearbox stages. found numerically by means of a two-stage transmission (Table 3). Figure 9c shows the engagement of the dog gear and dog. In thisFirst case, Stage the regeneration Second Stage system Total is functioning. Ratio In normal operation, the dog gear and dog are separate from each other. The dog is controlled by a hand level that pushes the 6:1 (120/20) 2.8:1 (56/20) 16.8:1 dog back and forth along the shaft. This simple system allows us to avoid parasitic losses of the regeneration pump, since this component is turned on only when required by the cyclist. 5. Electronic System Design PurdueTracer provides usersTable with3. Regeneration an electronic pump system gearbox for stages comprehensive. control, guidance, and recommendations, aiming to achieve seamless interaction between users and hydraulic systems, First Stage Second Stage Total Ratio and to improve the safety and intelligence of the vehicle. The connection of the electronic system 6:1 (120/20) 2.8:1 (56/20) 16.8: 1 is illustrated in Figure 10. An android application is utilized for user interaction. It sends out command output to Arduino via Bluetooth Low Energy (BLE), analyzes and displays real-time bicycle Energies 2018, 11, x FOR PEER REVIEW 15 of 24

5. Electronic System Design PurdueTracer provides users with an electronic system for comprehensive control, guidance, and recommendations, aiming to achieve seamless interaction between users and hydraulic systems, Energies 2018, 11, 305 15 of 24 and to improve the safety and intelligence of the vehicle. The connection of the electronic system is illustrated in Figure 10. An android application is utilized for user interaction. It sends out command output to Arduino via Bluetooth Low Energy (BLE), analyzes and displays real-time bicycle information, and generates personalized recommendations. The microcontroller Arduino is connected information, and generates personalized recommendations. The microcontroller Arduino is with the sensors and executive components on the bicycle. It reads the measurements and performs connected with the sensors and executive components on the bicycle. It reads the measurements and functionsperform tos instruct functions the to mode instruct and the gear mode ratio and change.gear ratio change.

FigureFigure 10. 10Overall. Overall connectionconnection of of e electroniclectronic system system..

5.1. Software 5.1. Software The software is developed based on the android application programming interface (API) and Thecan be software accessible is to developed users via smartphone based on the. As android shown in application Figure 11, the programming three major subsystems interface of (API) the and can beapplication accessible include: to users via smartphone. As shown in Figure 11, the three major subsystems of the application include:  The control subsystem for gear and mode change; • The controlThe display subsystem subsystem for to gear show and directed mode measured change; data from sensors, the analysis of bicycle and • The displayhuman performance, subsystem toand show recommendations directed measured for healthy data riding; from sensors, the analysis of bicycle and  The user-oriented design to enable comprehensive guidance. human performance, and recommendations for healthy riding; • The user-orientedThe control subsystem designto is enable designed comprehensive to facilitate hydraulic guidance. and mechanical system condition changes of PurdueTracer. The setup of the system includes the android application, Bluetooth Low TheEnergy, control microcontroller subsystem isArduino designed, and to sub facilitate circuits hydraulic for mode and and mechanical gear radio change. system The condition cellphone changes of PurdueTracer.receives the touch The setup input, ofmatches the system it into includes command the output android, and application, sends it to BLE,Bluetooth with data Low 0 Energy,–2 microcontrollerrepresenting Arduino,three types and of mode sub circuitschange (Pedaling, for mode Charge and gear, and radio Boost change.modes) and The 3– cellphone4 representing receives the touchgear ratioinput, adjustment. matches it The into Regeneration command output, Mode and is automatically sends it to BLE, executed with when data 0–2 users representing brake PurdueTracer and its electronic control will be explained in the next section. Arduino maps the three types of mode change (Pedaling, Charge, and Boost modes) and 3–4 representing gear ratio received command to corresponding functions and changes analog values or sends out impulse adjustment. The Regeneration Mode is automatically executed when users brake PurdueTracer and its electronic control will be explained in the next section. Arduino maps the received command to corresponding functions and changes analog values or sends out impulse signals at the pins for relays or optoisolators. Figure 12 illustrates the sub-circuits for controls. Both sub-circuits include a 12 V lithium battery, executive components (on–off valves for mode change and Shimano series), and a relay or optoisolator that can listen to the Arduino pin analog input to open or close the sub-circuit. The Shimano series is a commercialized product for gear ratio adjustment. To integrate the Shimano series into the electronic circuits, the shift switch is rebuilt. The cover and the mechanical button of the switch are abandoned, with the inner connection exposed for wire connection. Take one control option EnergiesEnergies 2018 2018, ,11 11, ,x x FOR FOR PEER PEER REVIEW REVIEW 1616 of of 24 24 signalssignals at at the the pins pins for for relays relays or or optoisolators. optoisolators. Figure Figure 12 12 illustra illustratestes the the sub sub--circuitscircuits for for controls. controls. Both Both subsub--circuitscircuits includeinclude aa 1212 VV lithiumlithium battery,battery, executiveexecutive componentscomponents (on(on––offoff valvesvalves forfor modemode changechange andand ShimanoShimano series)series), , andand aa relayrelay oror optoisolatoroptoisolator thatthat cancan listenlisten toto thethe ArduinoArduino pinpin analoganalog inputinput toto

Energiesopenopen oror2018 closeclose, 11, 305 the the s subub--circuit.circuit. The The Shimano Shimano series series is is aa commercializedcommercialized product product for for gear gear16 ratio ratioof 24 adjustment.adjustment. To To integrate integrate the the Shimano Shimano series series into into the the electronic electronic circuits, circuits, the the shift shift switch switch is is rebuilt. rebuilt. The The covercover and and the the mechanical mechanical button button of of the the switch switch are are abandoned, abandoned, with with the the inner inner connection connection exposed exposed for for wireaswire an connection. connection. example: when Take Take one usersone control control touch option option the “Gear as as an an Increase” example example: : button,when when users users an impulse touch touch the the signal “Gear “Gear will Increase” Increase” be generated button, button, anfroman impulse impulse Arduino signal signal and will sensedwill be be generated bygenerated the optoisolator, from from Arduino Arduino and theand and sub-circuit sensed sensed by by for the the gear optoisolator, optoisolator, change will and and close the the forsub sub a--circuit while,circuit fortofor generate gear gear change change the properwill will close close the for inputfor a a while, while, signal to to for generate generate Shimano the the Series proper proper and the the executes input input signal signal its function for for Shimano Shimano to mechanically Series Series and and executesadjustexecutes the itsits gear function function ratio. For toto mechanically themechanically touchscreen adjust adjust input thethe for geargear execution, ratio.ratio. ForFor there thethe are touchscreentouchscreen three buttons input input for modeforfor execution,execution, controls thereandthere two are are forthree three gear buttons buttons adjustment for for mode mode (Figure controls controls 13b). a andnd two two for for gear gear adjustment adjustment (Figure (Figure 13b). 13b).

ApplicationApplication

UserUser--orientedoriented ControlsControls DisplayDisplay DesignsDesigns

HydraulicHydraulic Recommendation MeasurementsMeasurements AnalysisAnalysis Recommendation GeolocationGeolocation ModesModes FunctionFunction

UserUser GearGear Ratio Ratio VelocityVelocity FlowFlow Rate Rate NavigationNavigation InformationInformation

Pump/HumanPump/Human Real-time WeatherWeather PressurePressure Real-time TorqueTorque RecommendationRecommendation PredictionPrediction

ContactContact & & HeartrateHeartrate EfficiencyEfficiency FAQFAQ

FigureFigure 11 11.11. .Hierarchy HierarchyHierarchy of of s softwaresoftwareoftware i implementationmplementationimplementation for for Android Android application.aapplicationpplication. .

Figure 12. Electronic circuits for controls. FigureFigure 12 12. .Electronic Electronic c circuitsircuits for for c controlsontrols. .

The display subsystem shows real-time measurement and analysis of bicycle performance, and gives recommendations to maintain healthy riding. Upon the activation of the application, the BLE automatically begins the scanning procedure and establishes the connection with the major Arduino Energies 2018, 11, x FOR PEER REVIEW 17 of 24

Energies 2018, 11, 305 17 of 24 The display subsystem shows real-time measurement and analysis of bicycle performance, and gives recommendations to maintain healthy riding. Upon the activation of the application, the BLE (identifiedautomatically by ID) begins placed the on scanning PurdueTracer. procedure The and direct establishes measurements the connection include velocitywith the of major the bicycle Arduino and human(identified pedaling, by ID) pressures placed on for PurdueTracer. main line and The accumulator, direct measure andments the heartrate include velocity of rider. of The the measured bicycle dataand are human stored pedaling, in order pressures in a buffer for tomain ensure line andthe continuity accumulator and, and stability the heartrate of data of transmission. rider. The PurdueTracermeasured data performs are stored real-time in order monitoring in a buffer of system to ensure performance, the continuity which and enables stability users of to data know transmission. PurdueTracer performs real-time monitoring of system performance, which enables their system operation condition and promotes safety by reducing the chance of unpredicted accidents users to know their system operation condition and promotes safety by reducing the chance of (Figure 13b). After acquiring system pressure and bicycle velocity, real-time calculations are performed unpredicted accidents (Figure 13b). After acquiring system pressure and bicycle velocity, real-time in the application to display real-time flow rate and torque. Simultaneously, PurdueTracer can provide calculations are performed in the application to display real-time flow rate and torque. riders with the instantaneous system efficiency by matching the operation conditions in the efficiency Simultaneously, PurdueTracer can provide riders with the instantaneous system efficiency by plots (Figure5). matching the operation conditions in the efficiency plots (Figure 5). InIn addition, addition, PurdueTracer PurdueTracer is equipped is equipped with with a recommendation a recommendation function function to avoid to excessiveavoid excessive exercise. Theexercise. maximal The heart maximal rate heart one should rate one have should is illustrated have is illustrated in Equation in Equation (13) for ( men13) for and men Equation and Equation (14) for women(14) for [22 women]: [22]: Hmax = 208.7 − 0.73 × A (13) 퐻푚푎푥 = 208.7 − 0.73 × 퐴 (13) Hmax = 208.1 − 0.77 × A. (14) 퐻푚푎푥 = 208.1 − 0.77 × 퐴. (14) As a user-oriented design, PurdueTracer provides personalized recommendations (Figure 13b). As a user-oriented design, PurdueTracer provides personalized recommendations (Figure 13b). Users can provide personal information at the time of sign-up on the log-in page (Figure 13e). Users can provide personal information at the time of sign-up on the log-in page (Figure 13e). The Theapplication application automatically automatically adjusts adjusts the the maxim maximalal heart heart rate rate criteria. criteria. DuringDuring the the pedaling, pedaling, PurdueTracer PurdueTracer constantlyconstantly monitors monitors the the rider’s rider’s heartrate heartrate and and comparescompares itit toto the maximum number. number. WhenWhen the the rider’s rider’s heartrateheartrate exceeds exceeds 90% 90% of of the the maximal maximal value,value, PurdueTracer will recommend recommend th thatat users users ride ride slowly slowly or or releaserelease stored stored energy energy to to aid aid with with the the riding riding processprocess ifif thethe accumulator has has enough enough pressure. pressure.

Figure 13. Android application design: (a) main page; (b) system monitoring, controls, and Figure 13. Android application design: (a) main page; (b) system monitoring, controls, and recommendation; recommendation; (c) weather prediction; (d) geolocation and navigation; (e) user log-in. (c) weather prediction; (d) geolocation and navigation; (e) user log-in. To further add to the user-oriented features, PurdueTracer includes design features that facilitate theTo usage further and addpromote to the the user-oriented comprehensiveness features, of the PurdueTracer system, as listed includes on the design main pag featurese of the that application facilitate the(Figure usage 13 anda). Users promote can theuse comprehensivenessthe geolocation function of the(Figure system, 13d) as to listedchoose on the theor mainpreferred page riding of the applicationlocation, record (Figure favorite 13a). Users places, can and use ge thet weather geolocation prediction functions (Figure (Figure 13c) 13 andd) tonavigation choose theor information preferred riding location, record favorite places, and get weather predictions (Figure 13c) and navigation information to reach the location. Frequently asked questions (FAQ) and contact information are implemented to help users. Energies 2018, 11, 305 18 of 24

5.2. Data Measurement Proper sensors are installed in the PurdueTracer as shown in Figure 10, to enable the functionalities of the application described in the previous section, but also to collect the data necessary to perform the model validation. Three different kinds of sensors (a total of ﬁve sensors) are installed to collect data regarding speed, hydraulic pressure, and user heartbeat.

• Velocity Sensor: Hall Proximity Switch Sensor

Two hall sensors were used. As shown in Figure 10, Sensor 8 was installed at the wheel to measure the translational speed of the vehicle; sensor 9 was used to determine the pedaling speed of the rider, which was installed on the gear. For this purpose, a small magnet was installed on the wheel. With the gear rotation, pulses are generated by the hall sensors and properly processed by the Arduino microcontroller to calculate the real RPM of the wheel and the gear. Speciﬁcally, both sensor 8 and sensor 9 have pins to receive and transmit the signals: a power pin (connected to a 5 V source), a ground pin, and a signal pin (connected to the analog pins on the Arduino microcontroller). The Arduino microcontroller was programmed to read the status of Analog pin 2 and Analog pin 3 continuously to catch the pulses generated by these hall sensors, which switch the signal from a low- to a high-level voltage signal every time the sensors approach the magnet. So the time interval between the two generated pulses represents the time spent on going forward the distance of one wheel circumference. The current speed can be calculated I as illustrated in Equation (15): p vmeasure = . (15) ∆t

• Pressure Sensor: Parker’s IQAN-SP 0–500 bar Pressure Sensor

Pressure sensors 6 and 7 in Figure 10 are placed in the hydraulic circuit to measure the pressure in the mainline connecting the pump and the motor, and to measure the accumulator pressure, respectively. Also, the pressure sensors have three pins to receive and transmit the signal (voltage). The signals generated by these pressure sensors were connected to proper analog pins of the Arduino microcontroller. The Arduino microcontroller would map voltages between 0 V and 5 V into integer values between 0 and 1023. To obtain the actual pressure, this integer value is ﬁrst converted into a voltage value: integer V = × 5 V. (16) out 1024 Then, according to the empirical relationship between hydraulic pressure and output voltage provided by sensor manufacturer (Parker), the value of the voltage signal has a linear relationship with the actual hydraulic pressure, as illustrated in Equation (13):

Vout = 0.5 + 0.008 × Pmeasure. (17)

• Heart Rate Sensor: Pulse Sensor Amped

A heart rate sensor (Figure 10) was installed on the handle bar for monitoring the health of the rider. It combines a simple optical heart rate sensor with amplification and noise cancellation circuitry, which make it fast and easy to get reliable pulse readings. The sensor consists of a bright green LED and light detector. Generally, the light is bright enough to pass through the finger and be detected by a light detector. When the heart pumps a pulse of blood through the blood vessels, the finger becomes more opaque to make less light reach the detector, which generates a different voltage signal to be received by the Arduino microcontroller. The Arduino would record the time interval between two different signals, which represents the time spent pumping one pulse of blood. So the BPM (beats per minute) could be calculated in real time. The pulse sensor also has three pins and the signal pin was connected to a corresponding analog pin 1 of the Arduino microcontroller. Energies 2018, 11, 305 19 of 24

Energies 2018, 11, x FOR PEER REVIEW 19 of 24 5.3. Regeneration Mode Implementation 5.3. Regeneration Mode Implementation The regeneration mode is set by the cyclist when he/she actuates the braking lever. As shown in Figure 14The, the regeneration regeneration mode controls is set by the the setting cyclist of when the proportionalhe/she actuates relief the braking valve through lever. As ashown proportional in lever.Figure The system14, the regeneration requires a micro-controller controls the setting and of the a separate proportional 9 V rechargeable relief valve through battery. a proportional The relationship betweenlever. the The input system voltage requires and a relief micro valve-controller pressure and agiven separate by the 9 V manufacturer rechargeable (Sunbattery. Hydraulics The relationship between the input voltage and relief valve pressure given by the manufacturer (Sun corp, Sarasota, FL, USA) is used to set the voltage level to the control valve. The pressure setting of Hydraulics corp, Sarasota, FL, USA) is used to set the voltage level to the control valve. The pressure the reliefsetting valve of the can relief be valve lowered can be by lowered increasing by incre theasing input the voltage input voltage signal, signal, which which is determined is determined by the proportionalby the proportional lever installed lever in installed the bike. in The the lever bike. signal The lever is used signal as input is used by as the input Arduino by the microcontroller. Arduino The microcontrollermicrocontroller. The has microcontroller to consider that has a higherto consider voltage that a input higher signal voltage from input the signal lever from (corresponding the lever to an increased(corresponding braking to command an increased by braking the cyclist) command corresponds by the cyclist) to a lower corresponds voltage tosignal a lower to be voltage sent to the reliefsignal valve. to be sent to the relief valve.

FigureFigure 14. 14.Circuit Circuit for r regenerationegeneration subsystem subsystem..

6. Results6. Results and and Model Model Validation Validation

6.1. Performance6.1. Performance Experiment Experiment OnceOnce the the PurdueTracer PurdueTracer was was built, built, experiments experiments were were performed performed to toverify verify the theproper proper functioning functioning and theand accuracythe accuracy of the of numericalthe numerical modeling modeling procedure procedure described described in in the the previous previous sections sections of of this this paper. paper. Table4 shows the measured performance, which can be compared with the numerical results in Table 4 shows the measured performance, which can be compared with the numerical results in TableTable2. 2. Improvements in the measuredTable performances 4. Measured inperformances. boost mode were expected due to the gear shifting (not present in the numerical model). Measured Performance in Pedaling Mode * Table 4. Measured performances. Bike speed 10.0 m/s

Main line pressureMeasured Performance 101 bar in Pedaling Mode * Bike speed 10.0 m/s Measured Performance in Boost Mode Main line pressure 101 bar Max speed Measured Performance 7.22 m/s in Boost Mode EfﬁciencyMax function speed (OF2)7.22 59.74 m/s Distance covered 240 m * Speed around 100–110 rpm. Energies 2018, 11, x FOR PEER REVIEW 20 of 24

Efficiency function (OF2) 59.74 Energies 2018, 11, x FORDistance PEER REVIEW covered 240 m 20 of 24 * Speed around 100–110 rpm. Efficiency function (OF2) 59.74 SignificantEnergies 2018 is, 11 the,Distance 305 comparison covered between the simulated model240 m and the real test of the20 Boost of 24 Mode. In the test shown in Figure 14, the accumulator* Speed around was 100 set–110 with rpm a. pre-charge of 55 bar. The accumulator was chargedImprovements to 150 bars with in the the measured hand performancespump, and then in boost the mode stored were energy expected was due released to the gear to drive the Significant is the comparison between the simulated model and the real test of the Boost Mode. bicycle. shifting (not present in the numerical model). In the test shown in Figure 14, the accumulator was set with a pre-charge of 55 bar. The accumulator As illustratedSigniﬁcant in isFigure the comparison 15, the accumulator between the simulated pressure model of experimental and the real test results of the Boost show Mode. consistency was chargedIn the test to shown 150 bars in Figure with 14the, the hand accumulator pump, and was then set with the asto pre-chargered energy of 55 was bar. released The accumulator to drive the withbicycle. thewas prediction charged toof 150 the bars model. with theSome hand inaccuracies pump, and thenare related the stored to energythe estimated was released resistance to drive terms of EquationtheAs (7), bicycle.illustrated but the inresults Figure are 15, satisfactory. the accumulator pressure of experimental results show consistency withFigure the predictionAs16 illustratedshows ofthe inthe Figureresults model. 15 of, Some the the accumulator inaccuraciesOF2 function pressure are related related of experimental to thethe estimated boost results mode. show resistance consistencyThe authorsterms of paid particularEquationwith attention the(7), predictionbut theto studying results of the are model. the satisfactory. effect Some of inaccuracies the accumulator are related pre to- thecharge estimated pressure resistance p0 on terms the achievable of OF2, whichEquationFigure was 16 (7), alsoshows but requ the the resultsired results for are satisfactory.ofthe the final OF2 NFPA function event related [17]. to The the modelboost mode.was particularly The authors usefulpaid for Figure 16 shows the results of the OF2 function related to the boost mode. The authors paid particular attention to studying the effect of the accumulator pre-charge pressure p0 on the achievable this study:particular as shown attention in Figure to studying 16, the effectbicycle of the reaches accumulator the maximum pre-charge pressureOF2 whenp on p the0 = achievable25 bar. OF2, which was also required for the final NFPA event [17]. The model was particularly0 useful for OF2, which was also required for the ﬁnal NFPA event [17]. The model was particularly useful for this this study: as shown in Figure 16, the bicycle reaches the maximum OF2 when p0 = 25 bar. study: as shown in Figure 16, the bicycle reaches the maximum OF2 when p0 = 25 bar.

Figure 15. The comparison of experimental results and model prediction for the change of FigureFigure 15. 15.TheThe comparison comparison of of experimental experimental results results and model and prediction model prediction for the change for of accumulator the change of accumulatoraccumulatorpressure pressure andpressure vehicle and and linear vehicle vehicle velocity linearlinear with velocity respect towith with time. respect respect to totime time. .

7070

6060 50

50[/]

[/] 2

OF 40

OF 40 30 30 20 20 20 30 40 50 60 20 30 po [bar]40 50 60 p [bar] Figure 16. Test to investigate the accumulatoro ’s optimal pre-charge pressure. FigureFigure 16. Test 16. Testto investigate to investigate the the accumulator accumulator’s’ optimals optimal pre-charge pre-charge pressure. pressure. 6.2. Competition Performance 6.2. CompetitionThe PurdueTracer Performance excelled at the NFPA final competition event held in April 2017 at Ames, Iowa. Three sub-competitions reflected the performance of the vehicle: The PurdueTracer excelled at the NFPA final competition event held in April 2017 at Ames, Iowa. Three sub-competitions reflected the performance of the vehicle:

Energies 2018, 11, 305 21 of 24

6.2. Competition Performance The PurdueTracer excelled at the NFPA ﬁnal competition event held in April 2017 at Ames, Iowa. Three sub-competitions reﬂected the performance of the vehicle:

• Sprint race: demonstrated the ability of the vehicle to move a short distance (about 200 m) quickly. • Efficiency challenge: demonstrated the ability of the vehicle to effectively store energy and efficiently use the stored energy to propel the unassisted vehicle. • Endurance challenge: demonstrated the reliability, safety, replicability, and durability of the fluid power system design and assembly.

A more detailed document on the rules of the competition can be found on the FPVC website [17]. Table5 shows a summary of the scores and judges’ evaluation of the Purdue team. Overall, the vehicle received quite a few positive comments from competition judges. The workmanship, innovative design of integrating a ﬂuid reservoir in the frame, and electronic control by smartphone app were the aspects highly favored by the judges.

Table 5. Summary of the scores at the NFPA ﬁnal competition (2017).

Podium Placement Overall Champion 1st Place Efﬁciency Challenge 1st Place Best Design 1st Place Reliability and Safety 1st Place Best Paper/Presentation 2nd Place Reliability/Durability Challenge 2nd Place Sprint Race 3rd Place Workmanship 3rd Place

7. Conclusions This paper presents the design methodology used to create the PurdueTracer: a light-duty hydraulic hybrid vehicle. PurdueTracer eliminates the classic chain-sprocket mechanical transmission, and implements a fluid power transmission system with electronic controls. The vehicle is capable of storing energy during charging phases or through regenerative braking. The hydraulic layout was developed to permit the vehicle to work in four modes selected by the user: normal pedaling, charging, boost, and regenerative braking. The selection of the hydraulic components and of the gearbox ratios was made through a numerical optimization on a lumped parameter model of the vehicle implemented in AMESim. Objective functions were defined to quantify the performance of the vehicle in pedaling and in accumulator boost operation. A lightweight aluminum chassis was designed to integrate the system reservoir and accommodate all the hydraulic components. An electronic control and data acquisition system was implemented to enable the different working modes and monitor the status of the vehicle and of the cyclist. The system is equipped with wireless controls, with an Arduino microcontroller collecting and analyzing sensor measurement, and an Android application performing advanced calculations and user-oriented functions for intuitive interactions. Measurements were performed to verify the initial model predictions. It was shown that the vehicle is capable of travelling 393 m using the 2 L accumulator, and can achieve a maximum velocity of 36 km/h during pedaling. This performance was achieved on the occasion of the 2017 National Fluid Power Association’s Vehicle Challenge. The PurdueTracer scored first in the following sections: Efficiency Challenge, Best Design, and Reliability and Safety, receiving the title of overall champion. Energies 2018, 11, 305 22 of 24

Acknowledgments: This work was partially founded by the National Fluid Power Association (NFPA), through the companies supporting the Pascal Society of the NFPA Education and Technology Foundation. The authors would also like to acknowledge the companies who contributed to this project with component donations: Casappa (gear pumps and motors); Sun Hydraulics, SunSource, and Eaton Corporation (hydraulic control valves); Steelhead Composites (accumulator); and Lube-Tech (hydraulic oil); Misumi (gears and bearings). Finally, the authors would like to thank Siemens PLM, for the in-kind use of the AMESim simulation software. Author Contributions: The paper is a joint effort of all the authors listed in the first page. All the authors, working as a team, contributed to the formulation, the design, the numerical analysis and the fabrication of the vehicle. In particular, Gianluca Marinaro is the leading author; he contributed to the hydraulic system designs and mechanical fabrications, with assistance from Zhengpu Chen and Yizhou Mao. Zhuangying Xu and Chenxi Li formulated and implemented the electronic data acquisition and control system; they also designed the smartphone app for the PurdueTracer. The whole project was supervised by Andrea Vacca, who also conceived the main ideas behind the hydraulic circuit design. All the authors participated to the NFPA Vehicle Challenge 2017 national competition (April 2017), representing the team of Purdue University. Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

Nomenclature

A user age, year 2 avehi acceleration of the vehicle, m/s Cx drag coefficient in longitudinal direction f coefficient of viscous friction, N/(m/s) Faero aerodynamic force, N Ff riction friction force, N Fslope slop force, N FW propelling force, N Hmax maximal user heart rate, bpm 2 Iw wheels inertia, kg·m L distance traveled by the vehicle, inch M cyclist and vehicle mass, lbs mcyclist cyclist mass, kg mvehi vehicle mass, kg n polytrophic exponent (ideal gas) nm motor shaft rotational speed, rpm np pump shaft rotational speed, rpm OF objective function of the optimization problem p wheel circumference, m p0 accumulator pre-charge pressure, bar p0,race accumulator pre-charge pressure during the competition, psi P0,reg initial accumulator pressure in regeneration mode, bar pacc nitrogen gas pressure in the accumulator, bar pmax accumulator max working pressure, bar Pmeasure measured pressure, bar pmin accumulator min working pressure, bar Qm flow needed for fluid motor speed, L/min Qp pump flow, L/min rm gearbox ratio of the motor rp gearbox ratio of the motor rW wheel radius, m 2 Sx vehicle active area for air resistance, m TW torque on the rear wheel, N·m Tm,e effective torque on the motor shaft, N·m Tp,e effective torque on the pump shaft, N·m v0,reg starting velocity in regeneration mode, m/s Energies 2018, 11, 305 23 of 24

3 Vacc accumulator volume, in 3 VD,m motor volumetric displacement, cm /rev 3 VD,p pump volumetric displacement, cm /rev 3 Vgas,acc nitrogen gas volume in the accumulator, m vmeasure vehicle velocity measured by electronic system, m/s Vout voltage signal output of pressure sensors, V vpedal vehicle velocity in pedaling mode, m/s Vvehi vehicle velocity, m/s Vwind wind velocity, m/s ∆pm pressure difference between motor inlet and outlet, bar ∆pp pressure difference between pump inlet and outlet, bar ∆t time interval for velocity measurement, s α road slope, ◦

ηhm,m motor hydro-mechanic efficiency ηhm,p pump hydro-mechanic efficiency ηv,m motor volumetric efficiency ηv,p pump volumetric efficiency µr rolling friction coefficients, 3 ρair air density, kg/m 3 ρair air density, kg/m

References

1. Chen, J.S. Energy efﬁciency comparison between hydraulic hybrid and hybrid electric vehicles. Energies 2015, 8, 4697–4723. [CrossRef] 2. Midgley, W.J.; Cebon, D. Comparison of regenerative braking technologies for heavy goods vehicles in urban environments. Proc. Inst. Mech. Eng. Part D J. Autom. Eng. 2012, 226, 957–970. [CrossRef] 3. Harriger, G.A. Hydraulic Drive System for Bicycles and the Like. U.S. Patent 4,546,990, 15 October 1985. 4. Schmidt, T.; Wilson, D.G. Human Power: Technical Journal of the IHPVA; IHPVA: San Luis Obispo, CA, USA, 2004; Volume 10, ISSN 0898-6908. 5. Neptune, R.R.; Kautz, S.A.; Hull, M.L. The effect of pedaling rate on coordination in cycling. J. Biomech. 1997, 30, 1051–1058. [CrossRef] 6. Chattin, J. Hydraulic Transmission for Bicycles. U.S. Patent 6,032,968, 7 March 2000. 7. Davey, J.A. Bicycles. U.S. Patent 4,290,621, 22 September 1981. 8. Brackett, D.C. Hydraulic Bicycle with Conjugate Drive Motors and Variable Stroke Crankshaft. U.S. Patent 5,938,224, 17 August 1999. 9. Chang, L.; Yao, J.F. The development of the electro-hydraulic driving bicycle. Appl. Mech. Mater. 2013, 373, 2132–2135. [CrossRef] 10. Yang, Y.H.; Zhong, X.Q. Design of Hydraulic Transmission Bicycle based on Top-down Method. Adv. Mater. Res. 2012, 468, 867–870. [CrossRef] 11. Amarantos, J.G. Hydraulic Powered Bicycle. U.S. Patent 4,087,105, 2 May 1978. 12. Swain, D.M.; Moore, J.Z.; Maurer, F.C.; Shih, A. Hydraulic Regenerative Braking for a Vehicle. U.S. Patent 7,992,948, 2 May 1978. 13. Lagwankar, K.D. Hydraulic regenerative system for bicycle. Int. J. Eng. Res. Appl. 2013, 3, 869–875. 14. Truong, D.Q.; Ahn, K.K.; Khoa, L.D.; Thinh, D.H. Development of a Smart Bicycle Based on a Hydrostatic Automatic Transmission. J. Adv. Mech. Des. Syst. Manuf. 2012, 6, 236–251. [CrossRef] 15. Wang, F.; Bissen, M.; Ward, W.; Stelson, K. Modeling and design of a hybrid bicycle with hydraulic transmission. Int. J. Fluid Power Syst. 2014, 8, 99–106. [CrossRef] 16. Foa, M.; Vacca, A.; Senatore, A. HYDROKART: A Human Powered Vehicle a Trasmissione Idrostatica; Oleodinamica e Pneumatica: Milan, Italy, 2017; ISSN 2421-4388. (In Italian) 17. National Fluid Power Association. Fluid Power Vehicle Challenge, Rules and Awards. Available online: http://nfpahub.com/fpc/vehicle-challenge/rules-and-awards-fpvc/ (accessed on 14 October 2017). Energies 2018, 11, 305 24 of 24

18. Wilson, D.G.; Papadopoulos, J. Bicycling Science, 3rd ed.; MIT Press: Cambridge, MA, USA, 2004; ISBN 9780262232371. 19. Navarro, K.F.; Gay, V.; Golliard, L.; Johnston, B.; Leijdekkers, P.; Vaughan, E.; Wang, X.; Williams, M.A. Social Cycle: What can a mobile app do to encourage cycling. In Proceedings of the 38th Annual IEEE Conference on Local Computer Networks—Workshops, Sydney, Australia, 21–24 October 2013; pp. 24–30. 20. Burke, E.R. High-Tech Cycling, 2nd ed.; Human Kinetics: Champion, IL, USA, 2002; ISBN 0-7360-4507-4. 21. Maestrelli, L.; Falsini, A. Bicycle optimization by means of an advanced gradient method algorithm. In Proceedings of the 2nd European HyperWorks Technology Conference, Strasbourg, France, 30 September–1 October 2008. 22. Tanaka, H.; Monahan, K.D.; Seals, D.R. Age-predicted maxima heart rate revisited. J. Am. Coll. Cariol. 2001, 37, 153–156. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).