Automotive Innovation

, Volume 2, Issue 1, pp 14–25 | Cite as

Novel Mechanical Interface Design for Automotive Starting Systems

  • Alvaro C. MichelottiEmail author
  • Jonny C. da Silva
  • Lauro C. Nicolazzi


Stop–start vehicles (SSVs) represent a potential alternative for improving internal combustion engine (ICE) efficiency. SSVs provide ICEs with the functionality of turning the engine off during traffic halts and restarting it without intervention by the driver. This strategy reduces fuel consumption, especially in dense urban traffic areas, and contributes to emissions reduction to meet green emissions targets. The most widely adopted SSV system has a mechanical interface to connect the electric starter motor to the ICE, which requires increased robustness compared with standard starting motors. This requirement allows the motor to withstand a higher number of engine start cycles compared with a standard starting motor. Nevertheless, it is a critical problem for wider adoption of SSVs. As SSV systems usually are based on the conventional starting system, its durability and noise remains a critical issue to be addressed by automakers. The typical pinion–ring gear interface uses intermittent gear meshing to form a transient coupling interface. The research reported here presents the development of an innovative mechanical interface for starting systems, called the permanent coupling (PC)-type interface, which reduces noise and increases durability compared with the existing design. The results obtained by a functional prototype of the PC-type mechanical interface confirm the feasibility of the proposed concept. The methodology is based on a product development process integrated with lumped-parameter modeling and virtual simulation aimed at reducing failures during prototype testing. The new mechanical interface was proven to be a good candidate for increasing the use of SSVs in the automotive market.


Starting systems Internal combustion engines Stop–start vehicles Mechanical interface Lumped-parameter simulation 



American Standard Code for Information Interchange


Carbon dioxide


Degree of freedom


Internal combustion engine


One-way clutch


Permanent coupling


Product development process


Parts per million


Stop–start vehicle


Transient coupling

1 Introduction

Critical environmental issues have highlighted the contribution of automobiles to the atmospheric pollutant emissions worldwide. According to the International Energy Agency [1], climate scientists noted that concentrations of carbon dioxide (CO2) in the atmosphere have increased significantly in the last century compared with levels in the pre-industrial era (about 280 ppm). It is estimated that the transport sector is responsible for about 22% of global CO2 emissions related to the impacts of climate change [1]. In a study by McKinsey & Co. [2], it can be seen that CO2 emissions from passenger cars can be reduced by 2.2 Gt per year by 2030.

Half of this goal could be met by improving the efficiency of internal combustion engines (ICEs) and their fuels. Biofuels and approaches that focus on driver education, as well as improving roads and infrastructure, could provide the remaining emissions reductions. Only 25% of the reduction would need to come from alternative technologies. ICEs require an external power source to generate the crankshaft initial rotation until it reaches autonomous operation. Vehicles equipped with ICEs have an auxiliary starting system to provide such functionality. The current configuration of ICE starters has been in production for over 100 years, as the first patent was granted in 1911 [3]. Theses motors have a mechanical interface comprising several components that make the connection between the shaft of the auxiliary starter motor to the ICE crankshaft.

Starting systems in the present research are classified into two types according to the mechanical interface: transient coupling (TC) type and permanent coupling (PC) type. In the TC-type mechanical interface, the starter motor has an intermittent gear that meshes with the pinion–ring gear located on the flywheel. At every engine startup, these gears must engage and disengage, causing premature wear and generating noise. In a PC-type mechanical interface, the gears are permanently meshed, which avoids gear noise and excessive impact wear during starting cycles.

Currently, starting system could contribute to reduced ICE emissions by an electronic management system that allows the ICE to automatically stop and start during short halts. Vehicles that adopt such strategy are commonly known as stop–start vehicles (SSVs), and they provide an effective way to reduce fuel consumption in dense urban traffic areas. Now, a second generation of SSV technology is under development, with potential fuel saving even under highway traffic conditions.

Depending on the emissions cycle considered, SSVs can reduce fuel consumption by 8–20% and produce a proportionate reduction in pollutant emissions. This is the main driving factor for the adoption of this technology [4].

As pointed out in the research by Han and Lee [5], most design activities involve the reuse of prior knowledge to solve new design problems. That is not the case in the present research, because to allow the development of new concepts, a truly innovative approach should be implemented. Different methods could be evaluated, such as design heuristics proposed by Yilmaz et al. [6], to allow the design to go beyond the reuse of existing knowledge.

Proposed research is focused on developing a disruptive innovation, as defined by Foster [7], due to fact of the existing “innovation gap” in SSV applications, as described below. This innovation gap is preventing SSV from being more widely adopted in the market, especially considering that the TC-type mechanical interface provides limited durability to meet customer requirements for SSVs (usually more than 300,000 cycles, compared with 30,000 for non-SSV systems).

Innovative PC-type concepts to date, such as the system proposed by Asada et al. [8], face limitations due to higher cost, as they rely upon a flywheel with an overrunning clutch located inside the engine block. This approach adds enormous design challenges during design validation. In addition, it can generate potential field failures due to the fact that the clutch needs a reliable dynamic sealing in the interface of the crankshaft, flywheel, overrunning clutch, and engine block [8].

This complexity compared to with the TC interface explains why new PC concepts for mechanical interfaces are not widely accepted by automakers. Nevertheless, several patents and patent applications exist, and these were reviewed during the literature search for this study.

A similar gap regarding technical limitations and economic feasibility is found in other proposed mechanical interface concepts, such as the concept developed by [9], in which a high number of components increase the acquisition cost and avoid its adoption by vehicle manufacturers.

Therefore, we propose an innovative PC-type mechanical interface with a simpler design that could overcome the existing technical complexity. The mechanical interface for the starting system was evaluated through dynamic simulation and experimental tests on a fully working prototype. This approach is similar to development approaches already pursued for other types of automotive systems [10, 11, 12].

This paper is organized as follows. In Sect. 2, the proposed methodology is described. In Sect. 3, we present system description. Section 4 shows the model validation based on experimental comparison. Finally, some conclusions of the research along with proposed future work are considered in Sect. 5.

2 Proposed Methodology

The research methodology adopted in the present work for development of an innovative product is composed of tasks in a structured product development process (PDP). The PDP applied herein follows the method developed by Pahl et al. [13], which has four design phases: (1) clarification of the task, (2) conceptual design, (3) preliminary design, and (4) detailed design. To clarify the proposed methodology, each phase is described in the following sections, along with the results obtained during the application of the proposed method to develop the novel mechanical interface.

2.1 Clarification of the Task

During task clarification, the main objective is to obtain design specifications for the new product, that is, the new mechanical interface. Several operating parameters of conventional starting systems and operating conditions required for SSVs are evaluated with this approach. Table 1 shows the customer requirements ranked by importance, as suggested by Hauser and Clausing [14]. In general, the “customer” is the automotive industry.
Table 1

Customer requirements for the mechanical interface





Improved SSV performance



Low weight



Fast and low noise/vibration ICE startup



Application in any type of ICE



Low cost



Installation without modification of ICE


1 = most important, 5 = least important

Following the proposed methodology, a multifunctional team developed a quality function deployment of the PC-type interface, defining a priority list of the design requirements, from both the customer’s point of view and technical feasibility based on available technology. For more information in relation to the new PC-type concept of mechanical interface, its global function and corresponding deployment into basic functions, see previous work done by Michelotti and Silva [15]. Design specifications, including the verification method and acceptance criteria, are listed in Table 2.
Table 2

Mechanical interface design specifications


Verification method

Acceptance criteria

Plug-and-Play design

Prototype test in ICE

Installation without changing ICE

Current envelope

Prototype test in ICE

Installation without changing ICE

High durability

Inherent design aspects

Estimated durability of the system

High starting torque

Max. torque capacity

Estimated torque capacity

Low overrunning torque

Overrunning torque

Overrunning torque < standard system

Low cost

Inherent design aspects

Estimated cost < 130% × std system

Shorter starting cycle

Inherent design aspects

Starting cycle time < standard system

2.2 Conceptual Design

During the conceptual design phase, a preliminary sketch of a newly proposed system was developed [15]. The new PC-type mechanical interface [16] has the subsystems identified in Fig. 1.
Fig. 1

Diagram of the PC-type mechanical interface

According to Fig. 1a, main subsystems are the torque and overrunning subsystem (1) and the impact absorption subsystem (2). Supplemental elements indicated in Fig. 1 are related to supporting components, such as the pinion (3), ring gear (4), flywheel (5), and bearing (6). Figure 1b shows the components of subsystem 1: return spring (7), pawl guide (8), and pawl (9); Fig. 1c shows the components of subsystem 2: arc-spring (10), spring guide (11), and spring end stops (12). Based on this structure, dynamic problems will be encountered in subsystems (1) and (2), as shown in Fig. 1.

In such subsystems, it is mandatory to ensure proper design to minimize the impact of the ratchet profile on the pawl during the coupling of the system for torque transmission (backlash effect). This coupling should occur with minimum noise and vibration, such as impact absorption. The proposed system provides a solution for such functionality and assures high durability for the coupling system.

Subsystem 1 in Fig. 1 must also provide zero overrunning torque during ICE stand-alone operation. In other words, during ICE autonomous operation, the new interface should have no contact between the pawl and ratchet profile in the ring gear. Total pawl decoupling with flywheel rotation must be correctly adjusted for this objective.

2.3 Preliminary Design

During the preliminary design phase, we focused on the analysis of the proposed system and the construction of a functional prototype for bench testing. Here, the preliminary design was accomplished with the aid of a lumped-parameter modeling approach. With this dynamic simulation, the proposed methodology emphasizes the benefits of the virtual development applied to innovative projects as a way to potentially reduce the occurrence of unexpected failures in the first prototype test. Two critical operational modes were evaluated with the aid of the virtual simulation, as described in more detail in the next sections.

2.4 Detailed Design

During this phase, the goal is to test a physical prototype. Proposed PC-type mechanical interface prototyping activities are presented, including manufacturing, assembly, and preliminary functional tests of the first physical prototype. As denoted by Jang and Schunn [17], late adoption of physical prototypes was a key characteristic of unsuccessful teams. Therefore, the approach followed herein was to build a first prototype as soon as possible to confirm the modeling and simulation results.

3 System Description and Results

This study aims to establish the critical parameters for the proper functioning of the new system. As highlighted in the previous section, the system should operate in such a way that the pawl remains fully open during ICE operation. Thus, it should operate in a zero overrunning torque condition, which is needed to meet the requirement for high durability.

Modeling in the present study was based on lumped parameters or 1D modeling. The dynamic system analysis is a function of only one single independent variable (time), and there is no additional spatial dependence (geometric, 2D, or 3D). We used the multiport method, which allows a technological representation of the system by defining modeling elements such as the spring, damper, inertia, and clutch [18]. This method creates a system representation that is more easily understood by product development teams in an industrial environment.

The subsystem model was implemented using the signal library of the commercial software AMESim (Advanced Modeling Environment for Simulations), which was chosen to evaluate the new concept. Further information on the simulation platform can be found in Silva [19].

3.1 Simulation of Mechanical Interface During Overrunning Operation

A critical aspect of pawl–ratchet systems already pointed out by Roach [20], and more recently by Kremer [21], relates to impact forces during overrunning operation, which generates unwanted noise. This noise can be understood according to what happens when someone rides a bicycle down a hill without using the pedals. In that case, if the bicycle model has a pawl–ratchet mechanism, the typical impact noise of the ratchet system is a well-known intermittent noise. In our example, the overrunning system allows the wheels of the bicycle to exceed their speed in relation to the rotational velocity of the pedals. The diagram of the dynamic model in Fig. 2 indicates each component as part of the developed model to evaluate the decoupling effect in the pawl–ratchet subsystem.
Fig. 2

Schematic diagram of the preliminary one-way clutch (OWC) system

Input data of the simulation are the flywheel angular velocity ωv. The pawl guide is attached to the flywheel, and it has a pin on which one end of the pawl can rotate. The pawl acts as a lever with angular displacement θ, which is assembled concentric to the ring gear ratchet profile. The inertia force, indicated in Fig. 3 as Fc, generated due to the flywheel movement, must exceed the return spring force Fs.
Fig. 3

Diagram to study the uncoupling of the pawl during overrunning

The mathematical model includes inertia elements that represent the kinetic energy stored in the system. In rotational systems, this energy is represented by the moment of inertia J. Thus, the flywheel rotating through its center fixed to the ICE crankshaft has a moment of inertia defined according to Eq. (1):
$$J = \int\limits_{\text{m}}^{{}} {r^{2} } {\text{d}}m$$
where r represents the distance of an infinitesimal element dm of the center axis of the flywheel. For translational elements, rigidity is represented by spring element k, whereas for rotational systems the equivalent is the torsional spring kt.

Two common types of damping elements are viscoelastic damping and friction damping. In the viscous damping, the damping force is proportional to the speed through the damper, acting in the opposite direction to that of the speed.

Dry friction force is also quite common between two surfaces of rigid bodies, such as Coulomb’s friction. Figure 3 shows a diagram used to develop the mathematical model which describes the pawl movement during the overrunning operation of the system.

The moment Mp in the pawl, which is required to decouple it from the ratchet profile, is given by Eq. (2):
$$M_{\text{P}} = \omega_{\text{v}}^{2}R_{\text{p}} R_{\text{v}} m \!\sin\!\theta$$
where ωv = angular velocity of the flywheel (rad/s2), Rp = position vector of the center of rotation of the pin to the center of mass of the pawl, Rv = position vector of the center of rotation of the flywheel to the center of rotation of the pin, θ = angle between Rv and Rp (radians), and m = mass of the pawl (kg).

The number of degrees of freedom (dofs) used in a mechanical system analysis is the number of kinematically independent coordinates required to describe the spatial motion of a particle in a TC-type system in a given instant of time. The number of dofs is reduced when the system is subject to movement restrictions. For instance, some joints used in mechanisms may have some constraints that reduce the dof of a given system. A prismatic joint, for example, allows only one dof.

In the free-body diagram of Fig. 4, the rotation of the flywheel ωv, which places the pawl in a rotating reference frame, is indicated where the rotational motion in the angle θ is represented by the linear displacement Δx of the pawl end.
Fig. 4

Free-body diagram of the OWC pawl motion

The counterclockwise rotation is generated by the inertia force acting on the pawl center of mass Cm, as opposed to the return spring force Fs and the actual pawl weight. A defined pawl position is assumed during flywheel rotation. This particular position was chosen because pawl weight in this location acts exactly in the opposite direction to the direction of inertia force.

This situation would be the most unfavorable condition for uncoupling in the one-way clutch (OWC) system, because the inertia force must overcome both the pawl weight and the return spring force to ensure no contact between the pawl and ratchet profile during the overrunning operation.

This is a dynamic problem with one dof, assuming that the pawl movement can be described only by its angle of rotation. The moment Mp generated by the flywheel rotation ωv in which the vane is fixed in Cr opposes the return spring force Fm.

The pawl can move by angle θ. Assuming also that the pawl rotation angle is small, the approximation error for a straight line is not significant for evaluating movement at the pawl tip. Therefore, in the proposed model, the angular displacement is simplified for a linear displacement Δx.

The effective force is a function of the radially directed centripetal acceleration of the flywheel rotation center, exactly opposing the inertia force required for the pawl rotational movement. The model neglects Coriolis acceleration, because movement at the pawl tip takes place over a short distance. Further, it is constrained by the arc defined by the pawl rotation center and its respective length. The model as an iconic representation is shown in Fig. 5, and it was experimentally validated as discussed in [22, 23].
Fig. 5

Dynamic model to study the pawl decoupling effect

Note that the pawl is modeled as a concentrated mass, connected by a lever arm to a pin or pivot. The pawl is also subject to an inertia force generated by the flywheel acceleration, being also subjected to a compression spring force. The ratchet profile is modeled as a cam profile.

Simulation model used a recorded ICE experimental rotational speed as input. An external data file from actual bench testing was input into a cam profile with cam follower model to represent the ratchet profile. The cam profile was defined by an external ASCII file containing geometric data. The model ignores the effect of pawl bounce on the ratchet profile. The angular displacement of the cam is defined by the differential equation shown in Eq. (3):
$$\frac{{{\text{d}}\theta_{\text{c}} }}{{{\text{d}}t}} = \frac{360}{60}\omega_{\text{c}}$$
where θc = angular displacement of the cam and ωc = rotary velocity of the cam.
The transform ratio is obtained by derivation of the cam profile with respect to the cam angle a, as defined by Eq. (4):
$$tf_{\text{ratio}} = \frac{1}{1000}\frac{360}{2\pi }\frac{{{\text{d}}x}}{{{\text{d}}a}}$$
The linear velocity vc and displacement lc of the cam follower are computed with Eqs. (5) and (6):
$$v_{\text{c}} = \frac{2\pi }{60}tf_{\text{ratio}}\,\omega_{\text{c}}$$
$$l_{\text{c}} = \frac{1}{1000}x\,\theta_{\text{c}}$$
Finally, the torque tc applied to the cam is computed based on cam force fc with Eq. (7):
$$t_{\text{c}} = tf_{\text{ratio}}\,f_{\text{c}}$$

The model also has an elastic contact between two bodies in linear motion to represent the nonlinearity present when the ratchet profile is not in contact with the pawl. The gap or clearance is calculated and a zero or negative gap implies contact. Therefore, if there is no contact, the two bodies move independently. When contact occurs, a contact force is applied to both bodies. This effect consists of a spring force and a damping force. To give continuity to this force, the damping coefficient is modified so that it is zero at first contact and approaches its full value asymptotically, achieving 95% of its full value according to the specified penetration.

Following the model description, a linear mechanical lever that multiplies the velocity and force as input by a mechanical ratio dependent on lengths L1 and L2 is defined to set the output force generated by the ratchet profile. Output force f2 is calculated based on input force f1 according to Eq. (8):
$$f_{2} = f_{1}\,\frac{{L_{1} }}{{L_{2} }}$$

A model representing 1D motion of a two-port mass under the action of two external forces and frictional forces is set to represent the pawl. The model returns the velocity, displacement, and acceleration of the mass.

Finally, a model of an ideal spring with two ports having force as output at both ports has spring compression calculated as an internal variable. The derivative of the spring force is calculated with Eq. (9):
$${\text{d}}f = k\,(v_{1} + v_{2} )$$
The compression of the spring is then calculated with Eq. (10):
$$x = \frac{{f_{1} }}{k}$$
Figure 6 shows the simulation result during the OWC system operation, as the ICE flywheel accelerates to a higher speed than the ring gear (in a PC-type interface), whose rotation is determined by the starter motor.
Fig. 6

Simulation of OWC pawl displacement

In Fig. 6, pawl displacement is described as a function of the ICE rotation. It is observed that the rotation in which the pawl loses contact with the ratchet profile corresponds to ωv = 625 rpm in the ICE flywheel. This condition assures proper engagement of the starting system until the ICE functions autonomously. In addition, it maintains the non-contact requirement of the pawl–ratchet system throughout ICE functioning, preventing the noise and wear that would reduce system durability.

3.2 Simulation of Mechanical Interface During Pawl–Ratchet Engagement

The second crucial dynamic problem for the new mechanical interface, already described by Kremer [21], is what happens during torque transmission through the pawl–ratchet system, where a certain initial rotation is required until the pawl engages with the next ratchet tooth.

This impact during engagement, besides affecting durability, generates excessive vibration and undesirable noise. Impact force generated during the torque transmission as a pulse, which relates to the variation of the pawl movement, is represented in Fig. 7 by starting torque Tp applied during the time interval Δt, as most critical condition. The maximum possible distance between pawl faces and ratchet profile is distance X, also indicated in Fig. 7.
Fig. 7

Diagram of the impact absorption system for PC-type mechanical interface

In the coupling system, an impact force of the ratchet profile is generated on the pawl end, indicated by Fc in Fig. 7. A baseline model to evaluate the system dynamic behavior during engagement without any impact absorption is shown in Fig. 8.
Fig. 8

Dynamic model of mechanical interface without impact absorption system

The model uses data provided by an external ASCII file, whose input is actual ICE speed to the model. Then, this speed is transformed into force by multiplying it by the flywheel radius and introducing an end stop model to define the connection between the force generated by the starting torque and the displacement of the pawl inside the pawl guide.

A model representing 1D motion of two bodies under the action of external forces was defined. In this case, it represented the pawl movement in relation to the ratchet profile. The initial pawl position should be set between lower and upper end stop positions. The lower the contact damping coefficient is, the more likely the contact at the end stop will result in a bounce.

For each damping ratio z, the contact damping coefficient Db is defined in terms of stiffness matrix K and mass matrix M by Eq. (11):
$$D_{\text{b}} = \frac{2z}{1000}\sqrt {{\mathbf{K}}{\mathbf{M}}}$$
where K = equivalent spring rate matrix (N/m) and M = equivalent mass matrix (kg).
Pawl relative displacement X1 and velocity V1 in relation to the ratchet profile displacement X2 and its velocity V2 are set according to Eqs. (12) and (13):
$$X_{\text{rel}} = X_{1} - X_{2}$$
$$V_{\text{rel}} = V_{1} - V_{2}$$
Contact forces (min, max) between the pawl and ratchet are calculated using Eqs. (14) and (15):
$$F_{\min} = \left\{ {\begin{array}{*{20}c} {Kb_{\min} (X_{\min} - X_{\text{rel}} ) - Db_{\min} \left( {1 - e^{{\frac{{ - (X_{\min} - X_{\text{rel}} )}}{{Pd_{\min} }}}} } \right)\,V_{\text{rel}} ,} &\quad {X_{\text{rel}} < X_{\min} } \\ {0,} &\quad {X_{\text{rel}} \ge X_{\min} } \\ \end{array} } \right\}$$
$$F_{\max} = \left\{ {\begin{array}{*{20}c} {Kb_{\max} \,(X_{\text{rel}} - X_{\max} ) - Db_{\max} \left( {1 - e^{{\frac{{ - (X_{\text{rel}} - X_{\hbox{max}})}}{{Pd_{\max} }}}} } \right)\,V_{\text{rel}} ,} &\quad {X_{\text{rel}} > X_{\max} } \\ {0,} &\quad {X_{\text{rel}} \le X_{\max} } \\ \end{array} } \right\}$$
The derivatives of the state variables V1, V2, X1, and X2 were set according to Eqs. (16)–(19):
$$\alpha_{1} = \frac{{{\text{d}}V_{1} }}{{{\text{d}}t}} = \frac{{F_{{{\text{sum}}1}} - F_{\text{dry}} }}{{m_{1} }}$$
$$\frac{{{\text{d}}X_{1} }}{{{\text{d}}t}} = V_{1}$$
$$\alpha_{2} = \frac{{{\text{d}}V_{2} }}{{{\text{d}}t}} = \frac{{F_{{{\text{sum}}2}} - F_{\text{dry}} }}{{m_{2} }}$$
$$\frac{{{\text{d}}X_{2} }}{{{\text{d}}t}} = V_{2}$$
Results of the PC-type mechanical interface without any impact absorption system, as can be seen in the simulated impacts in Fig. 9, indicate that the ratchet profile and pawl impact each other with a peak force of about 13 kN (approximately 2900 lbf). This value is extremely high and applied during a very rapid engagement. This energy amount leads to increased noise and vibration during system operation, which compromises its acceptance by potential customers. In addition, successive impacts could reduce the durability of the pawl–ratchet interface.
Fig. 9

Impact force on the pawl without the impact absorption system

As noticed from the dynamic behavior, the pawl–ratchet system should have some means to reduce the reaction force Fs. Thus, an impact absorption system was proposed to gradually reduce Fc, minimizing the noise generation and vibration that would occur if the system condition is based only on rigid pawl retention by means of the retaining pin.

When simulating the model with an impact absorption system, two arc-springs were included in the baseline model and simplified as linear and ideal mechanical springs. The derivative of the spring force for each spring is set by Eq. (20), and the model is depicted in Fig. 10:
$${\text{d}}f = k(v1 + v2)$$
Fig. 10

Dynamic model of mechanical interface with impact absorption system

The spring compression is calculated with Eq. (21):
$$x = \frac{f}{k}$$
The stiffness of each spring is computed with geometrical parameters in the model using the expression for helical spring by Eq. (22):
$$k = \frac{{Gd^{4} }}{{8D^{3} n_{\text{a}} }}$$
where G = material shear modulus (N/m2), na = number of active coils, D = spring diameter (mm), and d = wire diameter (mm).
The mathematical model represented in the iconic system shown in Fig. 10 can be understood with the aid of Fig. 11, which depicts a mass–spring system with two dofs.
Fig. 11

Mass–spring system with two dofs

Using Newton’s laws, we obtain the equations of motion for the system with Eqs. (23) and (24):
$$m_{1} \frac{{{\text{d}}^{2} x_{1} }}{{{\text{d}}t^{2} }} + (k_{1} + k_{2} )x_{1} - k_{2} x_{2} = F(t)$$
$$m_{2} \frac{{{\text{d}}^{2} x_{2} }}{{{\text{d}}t^{2} }} - k_{2} x_{1} + k_{2} x_{2} = 0$$
To solve this type of dynamic problem, it is usual to use a matrix representation. For a non-damping system, the motion equation is given by the expression presented in Eq. (25):
$${\mathbf{M}}\frac{{{\text{d}}^{2} x}}{{{\text{d}}t^{2} }} + {\mathbf{K}}x = {\mathbf{F}}({\mathbf{t}})$$
where x is the displacement vector, M is the mass matrix, K is the stiffness matrix, and F(t) is the force vector generated by the starting motor torque in a linear relationship with the ring gear diameter. For a system with two masses, as shown in Fig. 9, the motion equations in the matrix form are represented by Eq. (26):
$$\left[ {\begin{array}{*{20}c} {m_{1} } & 0 \\ 0 & {m_{2} } \\ \end{array} } \right]\frac{{{\text{d}}^{2} }}{{{\text{d}}t^{2} }}\left[ {\begin{array}{*{20}c} {x_{1} } \\ {x_{2} } \\ \end{array} } \right] + \left[ {\begin{array}{*{20}c} {k_{1} + k_{2} } & { - k_{2} } \\ { - k_{2} } & {k_{2} } \\ \end{array} } \right]\left[ {\begin{array}{*{20}c} {x_{1} } \\ {x_{2} } \\ \end{array} } \right] = \left[ {\begin{array}{*{20}c} {F(t)} \\ 0 \\ \end{array} } \right]$$

It is considered that the mass–spring system is linear. The adjustment of the arc-springs to absorb the starting impact is done by analyzing the impact force on the pawl or on the arc-springs. Reducing the impact force on these elements is the simulation objective, having as a constraint the component masses, which influence the total flywheel weight, in addition to the restriction of space available for the installation of the arc-springs and other components in the flywheel.

The simulated results of the model with an impact absorption system show that it is able to reduce the dynamic force to 1.3 kN (approximately 292 lbf), as can be observed in Fig. 12. As a result, the proposed impact absorption system is considered a proper design solution to be implemented in the physical prototype of the proposed mechanical interface, as presented in the next section.
Fig. 12

Impact force on the pawl with the impact absorption system

A simplified model, including flywheel inertia and other data, was developed for a preliminary evaluation of the new mechanical interface. The mathematical modeling included starter motor input torque, mechanical interface, flywheel inertia effect, and the ICE override after it starts functioning autonomously.

The preliminary lumped-parameter model is developed to predict the simulated ICE rotation curve during ICE startup with the new mechanical interface. The model presented in Fig. 13 shows, as we have already seen, the appropriate causality of the physical variables involved, as well as the mathematical models that describe the physical behavior of the elements involved in the analysis.
Fig. 13

Dynamic model of the ICE driven by the mechanical interface

The model shown in Fig. 13 has an input data curve representing electric motor input torque during ICE startup. These data were obtained from measurement taken during bench testing and are similar to those defined for installation and testing of the new mechanical interface prototype. Critical parameters, such as the ICE startup resistive torque, depend on several factors, such as number of cylinders, crankshaft inertia, and lubrication condition, among others.

The torque generated by the electric motor is transferred to the starter motor inertia model. Following a simplified representation of the mechanical interface, the pinion–ring gear coupling (gear ratio) is used to transform the pinion rotation to the ICE ring gear with a ratio of 12.1:1. Efficiency loss in the pinion–ring gear coupling has not been considered at this point.

The model comprises an ideal spring damper system. The spring model provides force as outputs at both ports. A model of a rotary load under the action of two external torques is applied with provisions for viscous friction, Coulomb friction, and static friction. It computes also the rotary velocity and the angular displacement in degrees.

The shaft speed ω is computed by integrating acceleration, and the angle θ is computed by integrating the shaft speed. A model of an ideal rotary mechanical gear with variable ratio multiplies the rotary velocity by the gear ratio α to compute the velocity output with an assumption of 100% mechanical efficiency. The torque input is also multiplied by this gear ratio to compute an output torque. Velocity output is set using Eq. (27):
$$\omega_{1} = \alpha\,\omega_{2}$$
The output torque is set using Eq. (28):
$$t_{2} = \alpha\,t_{1}$$

Simulated data were compared with the first prototype experimental results, as discussed and presented in Sect. 4.

3.3 Experimental Test Setup

An ICE test bench with a standard TC-type mechanical interface was used to install and evaluate the working prototype of the new concept, as shown in Fig. 14.
Fig. 14

ICE test bench having a standard TC-type mechanical interface

The conventional interface transmission ratio is 13.9:1. In the adaptation made to the new PC-type mechanical interface prototype, this ratio was reduced to 12.1:1. A gear with inner teeth profile was proposed to compensate for the difference between the gear ratios and maintain the same external diameter of the existing flywheel. Prototype dimensioning of the new PC-type mechanical interface included, but was not limited to, bearing press fit, new flywheel design with proper geometry of the OWC system, and arc-spring specifications for proper impact absorption system functioning. The physical prototype was installed in the test bench using the same fixture pattern and bolts of the existing ICE flywheel (Fig. 15).
Fig. 15

New PC-type mechanical interface installed in test bench.

Adapted from [24]

4 Modeling and Experiment Comparison

Performance of the new (PC-type) and existing (TC-type) mechanical interface is compared by checking experimental measurement of the effective transmission ratio TRe against the theoretical transmission ratio (TRf). This is the parameter adopted to define the time for the complete coupling of the pinion–ring gear system for full-start torque transmission. Reduced starting time reflects reduced noise during each starting cycle and reduced wear in the gear coupling.

The condition TRe < TRt, after the starting torque transmission phase occurs, indicates that the ICE is “overtaking,” or operating autonomously, and the starting system is in overrunning mode. After this condition, TRe is gradually reduced because the ICE rotation speed increases and the electric motor speed decreases because it is turned off.

Assumptions considered for the experimental data analysis are as follows: (1) The start cycle is initiated by the first greater-than-zero rotational speed value, indicating that the ignition key is in the “start” position. (2) The time interval between event #1 and effective pinion–ring gear coupling is based on coupling being signaled by TRe equaling TRt. (3) Proof of the overrunning system operation, indicating that the electric motor shaft is no longer coupled by the mechanical interface, is signaled by decreasing TRe.

The effective transmission ratio of the TC-type and PC-type interfaces was compared experimentally to have indication of prototype performance regarding its potential to reduce engine starting cycle time. Figure 16 indicates the conventional TC-type interface performance. The required time of 0.2 s, at which TRe first equals TRt, is considered as the effective pinion–ring gear coupling per rule (2) described above.
Fig. 16

Effective transmission ratio (TRe) of TC-type interface during ICE starting cycle

Following the experimental validation, the same rules were applied to assumption that was evaluated for the analysis of the PC-type interface experimental data. In this case, Fig. 17 indicates that for the PC-type interface, the time interval required for the permanent coupling between the pinion and the ring gear was decreased to only 0.05 s.
Fig. 17

Effective transmission ratio (TRe) of the PC-type interface during ICE starting cycle

The 0.05-s interval to obtain effective coupling can be explained by the influence of the impact absorption system during engagement. Experimental results obtained with the prototype confirm theoretical studies previously developed by Michelotti and Silva [23], where a lumped-parameter model was developed to evaluate TC-type and PC-type interfaces, regarding the time required for ICE startup.

Based on the dynamic 1D model of the new mechanical interface, along with experimental results for the ICE effective rotation, a theoretical–experimental comparison of the starting behavior with the new PC-type mechanical interface is presented in Fig. 18. It can be seen in Fig. 18 that the simulated and experimental curves have good agreement during the first part of the starting cycle, when the starter motor activates the ICE through the mechanical interface until the overtaking point.
Fig. 18

Comparison of simulation and experimental data for ICE startup with new mechanical interface

After starting the ICE operation, the virtual model is set to the initial condition of the startup cycle on the test bench. Usually, after ICE startup, the engine is accelerated for a certain time to ensure adequate lubrication for the test bench, because short ICE operating cycles can cause premature wear in the ICE. In the virtual model, the transition during ICE overtaking is faster, a characteristic that can be improved with minor adjustments in the resistive torque and friction forces of the engine model.

The improvement of the mechanical interface based on the virtual model, as well as the results of theoretical–experimental comparisons, shows potential for future development. It can be the basis for other models and experimental procedures beyond our current research scope, such as the experimental correlation with a range of different ICE types. This can both assist with the previous analysis and promote additional improvements to the mechanical interface concept for different applications and specific customer requirements.

5 Conclusions

A novel mechanical interface for SSVs was successfully developed as a potential alternative to evolve the state of the art in starting systems for ICEs. Tests performed in a functional prototype confirmed the possibility of overcoming obstacles to improvement of the current SSV systems. The results pointed out some limitations of both TC-type and PC-type interfaces to meet the increased demands of SSVs. The creative effort applied in the research for the development of a new dynamic coupling system demonstrates how it is possible to expand current possibilities in design, manufacturing, and testing of viable alternatives for PC-type interfaces.

Feasibility studies must focus on both technical and economic analyses to be effective. The technical benefits of the novel mechanical interface were noticed during prototype testing, as indicated by the reduction of the starting cycle time from 0.2 to 0.05 s in the test bench results. Another benefit was noticed during engine normal operation, as the new PC-type interface made no contact between the ring gear and flywheel, assuring the potential for achieving very high durability requirements, as no wear is expected during normal engine operation. Ultimately, having no intermittent meshing between the pinion and ring gear assures that the gear profile will not be excessively damaged during starting cycles, allowing a long system lifetime. With the technical and cost-driven enhancements achieved, based on a simplified design compared with the current PC-type mechanical interfaces, the proposed novel design has a great potential to be adopted in future SSVs.

Future work involves further validation of the new design in an actual vehicle application. Additionally, simulation models can be expanded to represent different types of ICE applications, such as automatic transmission interface with ICE and heavy-duty engines. In addition, refined modeling can simulate a wider set of variables of interest, including complex contact modeling and comprehensive simulation of the ICE resistive torque and vibration analysis during the starting cycle.



The authors would like to thank ZEN S.A. Industria Metalurgica ( for sponsoring and funding the project. In addition, special acknowledgement is given to the Federal University of Santa Catarina, Florianopolis Campus (Brazil), as this paper is part of a doctoral thesis research done by the first author, with co-authors as advisors.


  1. 1.
    International energy agency: CO2 emissions from fuel combustion. IEA Statistics (2015)Google Scholar
  2. 2.
    Mckinsey & Company: Roads toward a low-carbon future—reducing CO2 emissions from passenger vehicles in the global road transportation system (2009)Google Scholar
  3. 3.
    Kettering, C. F. (inventor); Dayton Engineering Lab Co., (assignee): Engine-starting device. US Patent 1,150,523. EUA: USPTO (1915)Google Scholar
  4. 4.
    Navigant Research Inc.: Stop–Start vehicles. (2016). Accessed 25 June 2016
  5. 5.
    Han, Y., Lee, K.: A case-based framework for reuse of previous design concepts in conceptual synthesis of mechanisms. J. Comput. Ind. 57, 305–318 (2006)CrossRefGoogle Scholar
  6. 6.
    Yilmaz, S., Seifert, C., Daly, S., et al.: Design heuristics in innovative products. ASME. J. Mech. Des. (2016). Google Scholar
  7. 7.
    Foster, R.: Innovation: the attacker’s advantage. Summit Books (1988)Google Scholar
  8. 8.
    Asada, T., et al.: inventor; Toyota Jidosha Kabushiki Kaisha, assignee. Starting Apparatus. US Patent 7,472,672. EUA: USPTO (2009)Google Scholar
  9. 9.
    Lindemann, P., et al.: Torque converters: launching over new challenges. In: Schaeffler Symposium (2010)Google Scholar
  10. 10.
    Li, L., Gong, Y., Deng, J., et al.: CO2 reduction request and future high-efficiency zero-emission argon power cycle engine. Automot. Innov. (2018). Google Scholar
  11. 11.
    Huang, K., Yan, Y., Li, B., et al.: A novel design of thermoelectric generator for automotive waste heat recovery. Automot. Innov. (2018). Google Scholar
  12. 12.
    Yang, C., Yang, M., Liu, T.: Design and analysis of a novel centrifugal braking device for a mechanical antilock braking system. ASME. J. Mech. Des. 2, 1 (2015). MathSciNetGoogle Scholar
  13. 13.
    Pahl, G.: Engineering Design: A Systematic Approach, 3 edn. Springer, Berlin. ISBN 9781846283185 (2007)Google Scholar
  14. 14.
    Hauser, J.R., Clausing, D.P.: The house of quality. (1988). Accessed 10 Oct 2016. Harvard Business Review, pp. 63–73
  15. 15.
    Michelotti, A.C., Silva, J.C.: Design innovation in dynamic coupling of starting system for internal combustion engines. J. Braz. Soc. Mech. Sci. Eng. 38(1), 177–188 (2016)CrossRefGoogle Scholar
  16. 16.
    Canto Michelotti, A. (inventor), ZEN, S.A.: Industria Metalurgica (assignee): Decoupler with free wheel system and vibration damping and one-way clutch with free wheel system. US Patent 9,605,743. EUA: USPTO (2017)Google Scholar
  17. 17.
    Jang, J., Schunn, C.D.: Physical design tools support and hinder innovative engineering design. ASME. J. MEch Des. (2012). Google Scholar
  18. 18.
    Lebrun, M., Richards, C.: How to create Good Models without writing a single line of code. 5th Scandinavian Intl. Conference on Fluid Power, Linkoping, Sweden (1997)Google Scholar
  19. 19.
    Silva, J.C.: Virtual environment for dynamic modeling of multi-domain systems. 18th International Congress of Mechanical Engineering. Ouro Preto, Brasil (2005)Google Scholar
  20. 20.
    Roach, G.M.: An investigation of compliant over-running ratchet and pawl clutches. All Theses and Dissertations. Paper 55, (2003). Access 30 Jan 2016
  21. 21.
    Kremer, J. M.: Design practices: passenger car automatic transmissions (AE-29). Cap.7—one-way clutches”, 768 p. SAE International, Warrendale, USA. ISBN 9780768011258 (2012)Google Scholar
  22. 22.
    Michelotti, A.C., Silva, J.C.: 1-D Model development to study the dynamic behavior of the mechanical diode configuration applied to one-way clutch (OWC). XVIII SAE BRASIL Congress (2009)Google Scholar
  23. 23.
    Michelotti, A.C., Silva, J.C.: Dynamic model of the roller float condition in one-way clutch (OWC) applied to motorcycle engine starters. 20th International Congress of Mechanical Engineering—COBEM, ABCM (2009)Google Scholar
  24. 24.
    Michelotti, A.C., Silva, J.C.: Dynamic simulation of an innovative permanently engaged mechanism for starting systems. 23rd International Congress of Mechanical Engineering—COBEM (2015)Google Scholar

Copyright information

© Society of Automotive Engineers of China (SAE-China) 2019

Authors and Affiliations

  1. 1.Research and Development DepartmentZen S.A. Industria MetalurgicaBrusqueBrazil
  2. 2.CTC-EMC-UFSCUniversidade Federal de Santa Catarina (UFSC)FlorianópolisBrazil

Personalised recommendations