About this Journal Submit a Manuscript Table of Contents
Advances in Mechanical Engineering
Volume 2013 (2013), Article ID 596340, 10 pages
Research Article

Biomechanical Response and Behavior of Users under Emergency Buffer Crash

1Design and Manufacturing Department, EINA, University of Zaragoza, C/ Maria de Luna s/n, 50017 Zaragoza, Spain
2Mechanical Department, EINA, University of Zaragoza, C/ Maria de Luna s/n, 50017 Zaragoza, Spain

Received 26 June 2013; Revised 22 August 2013; Accepted 24 August 2013

Academic Editor: Magd Abdel Wahab

Copyright © 2013 R. Miralbes et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


This paper aims to study the biomechanical effects on elevator users and the injuries sustained should an elevator crash happen. The analysis will focus on buffer impact, signaling that the earlier mentioned buffer is usually located at the bottom of the pit. In order to carry out this analysis, a numerical technique based on finite element method will be used, while elevator users will be simulated by means of automotive dummies. Two crash factors will be studied, namely, location of dummy and fall velocity. The analysis criteria will be damages sustained by the dummy, based on biomechanical index such as HIC, CSI, forces, and accelerations.

1. Introduction

Safety in elevators has been a major issue in the elevator industry since the beginning of such transport systems. Furthermore, there is a high level of social awareness linked to elevator accidents that has brought about a lot of investigation into them. There are two main types of accidents, those involving car free falling [1] and car ceiling crashes due to the misdetection of top floor.

This paper will focus on the particular case of an elevator car free fall crash. In that case the elevator would crash into those pit polyurethane buffers, henceforward called puffers. There are two pit buffers types, those falling under the energy dissipation category, and those of under the energy accumulation one. Polyurethane buffers belong to the energy accumulation type, as well as those of steel spring type, whereas oil buffers fall under the energy dissipation category and are not as popular as polyurethane or spring based.

If an elevator car free fall should happen, three systems might stop the elevator. Firstly, the elevator brake should stop the engine. Were this to fail, and the brake did not stop the engine and the ropes, the overspeed governor should make the safety gear stop the elevator. In the unlikely case of a safety gear failure or an overspeed governor malfunction, the pit puffers [2] would absorb the kinetic energy of the car, turning such energy into strain energy due to the deformation of the puffers [3].

This paper will focus on the third situation of the elevator crash. This is when the elevator will crash into the buffers located at the bottom of the pit. This free fall would occur if the brake did not stop the car and if the safety gear was not activated by the overspeed governor.

Polyurethane buffers, puffers, turn kinetic energy into strain energy, thereby causing an attenuated collision between the car and buffer, leading to a soft and smooth landing of the car while avoiding serious accidents. These buffers have a higher energy absorption capacity compared to the steel spring buffers.

Two crash variations are presented in this paper, that is, location of dummy and fall velocity. The analysis criteria used are damages on dummy, based on biomechanical indexes such as head injury criteria (HIC), chest severity index (CSI), forces, and accelerations.

For the numerical calculations, a crash test dummy will be used. This dummy allows monitoring those main biomechanical indexes such as HIC and CSI. Accelerations and forces can be calculated as well by means of virtual accelerometers located in the dummy. These measures will allow damages and injuries sustained on the dummy due to the crash to be calculated.

Several variants will be calculated by changing both dummy location and considering various impact velocities.

2. FEM Models

In order to assess the damages sustained, the best option consists in making use of a crash test dummy placed inside an elevator while there is a crashing of buffers. However, this option has been ruled out as being too expensive.

The option chosen has been resorting to numerical techniques which allow researchers to simulate the real l conditions encountered while cutting down on cost and time.

This research will be performed by means of an effective and efficient use of a numerical tool such as the finite element method (FEM).

The main FEM code utilized for the calculations is the LS-DYNA code. This code is widely used for automotive crash and allows the use of an extensive family of crash test dummies. All dummies are available, and some of them are even free to use and widely correlated to actual test dummies. While LS-DYNA has been mainly used to calculate the crash, two other codes (ABAQUS and PATRAN) were used to model and mesh both elevator car and buffers based on a generic elevator car and the buffer model. Both models were calculated in order to check static and dynamic behavior, weight, structural integrity, and suitability of mesh in terms of stresses and calculation time.

The elevator car used is a four-people car, (Figure 1) 2.45 meters high, 1.14 meters deep, and 2.18 meters wide. Any non-structural parts such as mirror, decorations, and lights were removed and therefore not modeled nor meshed. Usually the car is made of steel plates with a relatively small thickness. That allows modeling the car with shell elements. A reduced integration was selected and a fine mesh consisting of about 180,000 elements and 190,000 nodes was necessary in order to model the car (Table 1 shows it properties).

Table 1: Stress-strain polyurethane behavior.
Figure 1: Elevator car finite element model.

Puffers are made of high-density polyurethane suitable for crash conditions. These puffers (Figure 2) belong to the latest generation of impact buffers and are covered with an aluminum sheet 1 mm thick. This part was necessarily modeled with solid brick elements for the polyurethane and shell elements for the cover sheet.

Figure 2: Puffer finite element model.

Once both car and buffers were modeled, a Hybrid III dummy model was placed into the car. Integration and positioning was done in LS-DYNA. Figure 3 shows the global model with car, buffers, and dummy fully integrated and ready to crash.

Figure 3: Global finite element model.

The elevator car was mainly made of steel. Both St-52 and St-44 were used for the car. The floor was also made of plywood. Plywood was simulated as well by means of solid brick elements due to its thickness. The puffer consisted of two integrated parts. The solid meshed part was polyurethane, whereas the other was the aluminum sheet cover, made of 5088 alloy. Steel and aluminum were modeled as elastic-plastic materials, following a stress-strain curve, using the LS-DYNA piecewise linear plastic option [4]. Plywood was simulated by means of the wood material option of LS-DYNA [4]. Scottish pine wood properties were selected for this particular elevator car floor. No failure was expected on the pine wood floor, while some plastic strain should occur on steel parts. High strain was expected on the puffer; therefore crushable foam with hardening was used to simulate polyurethane behavior. Properties for the polyurethane foam were as follows: Youngs modulus: 0.10917 MPa,C. Poisson: 0.3,density: 2.85*10−2 kg/mm3.

3. Load Cases and Boundary Conditions

Nine load cases were calculated. These load cases result from the combination of three free fall heights and three dummy locations inside the car. Therefore the influence of both factors will be assessed.

Heights correspond to free falls from the second floor (6 meters high), from the seventh floor (20 meters high), and from the ground floor. Ground floor free fall is considered as a fall at 2.5 m/s speed which corresponds to a high elevator velocity impact. Velocity is considered 2.5 m/s for the other load cases.

Numerical simulation will be carried out by means of dynamic explicit finite element formulation, which is the most suitable calculation method for highly dynamic impacts [4].

In order to reduce and optimize computing time, a crash is considered initiated when the car is about to hit the buffers. Calculation time lasts until the necessary time required to turn kinetic energy into some other types of energies is consumed. Kinetic energy will turn mainly into strain energy and some friction energy.

Therefore, the load cases begin with the car located extremely close to the buffers and bearing an initial velocity () calculated by means of (1), where is the elevator maximum rated velocity (2.5 m/s currently), is gravity and is free fall height. Free fall height will be determined by multiplying 3 meters by the number of floors considered for the load case. Therefore, the crash velocities would be 2.5 m/s, 11.1 m/s, and 20.4 m/s. These velocities are introduced into calculation as initial velocity of car. The dummy has the same velocity owing to gravity and contact boundary conditions between the whole body and the car. It is not necessary to introduce such velocity into the dummy, as the elevator car forces the dummy to travel at its own speed, which is the same as that of the elevator. Gravity acceleration must be introduced as well.

Boundary conditions are as follows: full autocontact (LS-DYNA automatic_single_surface) for all parts. This means that the elevator car and buffers, the dummy, and the car, and even all parts can enter into contact with one another if necessary. A static and dynamic friction of 0.3 has been considered.

As the elevator car moves along the guide rails, rail brackets are restrained as shown in Figure 4. Puffers are usually attached to a concrete cube at the bottom of the pit. Therefore the bottom part of the puffer is fully restrained. A concrete cube is modeled as a rigid wall, in order to simplify calculations. The puffers must stop the car before it impacts the floor, so this assumption should be suitable for the purpose of the paper. As concrete would always be under compression conditions, this assumption seems to be reasonable.

Figure 4: Boundary conditions.

The passenger is located on three different positions in order to calculate three load cases. These locations will be the centere, the lateral, and the corner. These three locations should allow knowing the influence which the location has on damages and injuries on impact. Figure 5 shows dummy locations for the load cases.

Figure 5: Dummy locations inside the car: center, lateral, and corner.

4. Dummy

There is a wide range of numerical dummies at our disposal. There are some companies offering dummies for many types of calculations and simulations, and these companies even offer the crash test dummy as well. The dummy is supposed to portray human behavior under different situations, and the crash test is one of the most typical ones [5]. While the crash test dummy is the major issue on dummies, dummies are widely used for many applications, including automotive, aircraft, and railway crashes among others and not necessarily on crash tests but to test accelerations or forces in somewhat risky situations. At present no dummy crash test has been carried out in the field of elevators [5].

Crash test dummies have been developed mainly for the automotive sector. The range of dummies is quite wide and includes models such as the Hybrid III (frontal crash seated dummy), FAT ES-2, ES-2re, FTSS SID-IIs, FAT EuroSID-1, FAT USSID, SIDHIII, for lateral crash or back crash (Bio RID II). The choice of dummy will depend upon standard or load case.

As there is no dummy specifically designed for the field of elevators, the choice will depend on three boundary conditions of the proposed test and their availability.

For this elevator crash, the most suitable model would be one used for frontal crash, rather than lateral or back crash. The Hybrid III dummy will be the chosen model. This dummy is currently used to simulate frontal crash under Euro NCAP standards. The HYBRID III 50th Percentile crash test dummy, representing the average adult male, is the most widely used dummy for frontal crash and automotive safety restraint testing. Originally, the Hybrid III 50th male was developed by General Motors for vehicle safety purposes. Over the years, improvements have been made to the dummy to make it more humanlike. It is usually considered for generic testing. For instance, this dummy has been considered for the UNE-EN_135000 standard regarding biker protection against road lateral protective walls.

Hybrid III percentiles, such as male 50th percentile (standard male, 168 cm height, and 77 kg weight), male 90th percentile (tall male, 188 cm height, and 100 kg weight), female 50th percentile (small female, 152 cm height, and 50 kg weight), kids six years old (21 kg weight), kids three years old (15 kg weight), and kids 18 months old, are available.

This dummy is usually in a seated position when used for automotive crashes such as the euron CAP frontal crash. Therefore, the dummy had to be modified to reach an upright position. Both neck and pelvis had to be tuned to match the standing position. The neck is easily modified by means of commands, whereas the pelvis had to be modified by means of a pelvis kit following the standard UNE 135900.

This is the reason for the somewhat weird external appearance of the dummy shown in Figures 4 and 6, although it does not affect the dummy’s capacity to measure accelerations and forces.

Figure 6: Hybrid III 50th percentile.

Dummy Hybrid III consists of the following parts: Head: made of aluminum and rubber. It is equipped with triple-axis accelerometers measuring brain accelerations. Neck: there are sensors measuring pit, pitch, forces, and accelerations. Arms: the arms are not too restrained due to their nature, therefore damages are not usual. Because of that, arms are not tracked by sensors. A visual checkout is performed. Chest: the ribs are tracked in order to asses chest crushing. Crushing is measured by means of displacements and forces. Femur and thighs: there are sensors between pelvis and knees. On the pelvis femur joint, sensors also measure forces and accelerations. Calf: compression, torsion, pit, pitch, shear forces at tibia and fibula are measured. Ankles and feet: sensors measure pit, pitch, and twist. Forces are measured. The current description shows dummy behavior no matter which crash test dummy or FEM model may be used. Of course, the dummy used is fitted with sensors and accelerometers, and FEM model just locates a node at the area where forces must be measured.

4.1. Load Case 0: Dummy Placement Area

Load Case zero is the initial dummy placement before any crash calculation. The dummy is located and placed horizontally and vertically as shown in Figure 5. Once correctly located, an initial gravity force is applied to the dummy for 3 seconds. The elevator is fully restrained at buffer impact zones. This load case allows the dummy to contact the floor.

There is no initial velocity at all, and the system is ready to proceed with the crash load case.

4.2. Load Case 1: Elevator Car Buffer Crash

This load case is generated once the dummy is placed by means of load case 0. Load case boundary conditions are as those previously specified, as well as the initial velocity for both elevator car and dummy, while the velocity on car would be enough as the gravity attaches the dummy to the floor. Initial velocity is 2.5 m/s, 11.1 m/s, or 20.4 m/s depending on load case.

The calculation time is enough to allow kinetic energy to be zero if that situation happens. Calculation time is 1.5 seconds. Later on, initial kinetic energy considerations, such as kinetic energy turning into strain energy, mainly fail to happen. This is due to the car and dummy rebounding after the crash. Therefore, it was required to measure vertical velocity and acceleration in order to take a decision on when to stop calculations. Of course, initial calculations required a longer time in order to evaluate how the model behaves.

Kinetic initial energy is considered as follows:

5. Biomechanical Indexes and Injuries

Biomechanical indexes are related to damages and injuries; therefore a basic knowledge on such fields is necessary in order to assess damages and injuries.

Hybrid III allows assessing the damages at head, neck, thorax, pelvis femur and lower extremities. Those zones are designed according to experimental data coming from testing performed on corpses. That is the best way to correlate dummy fidelity to human body response to a crash as stated by several authors in different studies (head [5], neck [6], thorax [7], pelvis [8], femur [9] and lower extremities [10]), on Hybrid III [11]. These authors provide the main criteria to analyze damages and injuries by means of biomechanical indexes calculated from dummy data.

Therefore, the main biomechanical indexes to assess damages and injuries are as follows.

Head. Main damage is due to decelerations on impact. Peak values are important, but sustained acceleration values are even more important. The available criteria for the head are the “head injure criterion” (HIC), defined as follows. where and are two time points during impact and is the acceleration considered (’s). HIC36 and HIC15 are defined as follows: where  ms like HIC15, HIC36 has been selected as it is more restrictive. The considered limits are: damage level I: 600 and damage level II: 1000 these limits correspond to a 2% and a 15%, respectively, of a permanent brain damage.

Neck. The following values are assessed (Figure 7). Neck flexion moment (MF < 190 Nm). Neck extension moment (ME < 57 Nm). Axial tensile neck loading (Fz, Figure 1 of [10]). Axial compressive neck loading (Fz, Figure 2 of [10]). Fore/aft shear force (Figure 3 of [10]).

Figure 7: Forces, accelerations, and moments in the neck/head.

Thorax. The following values are assessed.Sternum deflection (<60 mm). Sternum deflection rate (<8 m/s).Viscous criterion (<1 m/s). Thoracic spine acceleration (<60 ’s in less than 3 ms.). Chest severity index (CSI, <700).

Femur. The following values are assessed.Axial Compressive femur force (Figure 5 of [10], 9070 N).Relative translation of femur and tibia at knee joint (<15 mm).Combined bending and axial compressive loading of leg (<1). Medial and lateral tibia plateau compressive forces (<4000 N).

Lower Extremities. Medial and lateral ankle compressive forces (<4000 N).

All these parameters are available within the numerical Hybrid III dummy except for the thorax viscous criterion; there is a relative translation of femur and tibia at knee joint and the combined bending and axial compressive loading of leg.

It is worth mentioning that below the pelvis, damages or injuries should not be fatal, no matter how severe the injuries. Damages and injuries on neck, head, and thorax could cause many severe problems, even leading to death.

6. Results

Figures 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, and 20 show results for the three dummy locations and velocities. Biomechanical safety levels are analyzed for the biomechanical indexes previously shown.

Figure 8: HIC 15/36.
Figure 9: Axial tensile neck loading.
Figure 10: Axial compressive neck loading.
Figure 11: Fore/aft shear force.
Figure 12: Sternum deflection.
Figure 13: Thoracic spine acceleration.
Figure 14: Chest severity index.
Figure 15: Axial compressive right femur force.
Figure 16: Axial compressive left femur force.
Figure 17: Medial and lateral tibia right plateau compressive forces.
Figure 18: Medial and lateral tibia left plateau compressive forces.
Figure 19: Medial and lateral right ankle compressive forces.
Figure 20: Medial and lateral left ankle compressive forces.

Figures 21 and 22 graphically show dummy behavior at different velocities and locations.

Figure 21: Evolution of calculation for a center located dummy at 20 m/s speed crash. Frames are every 1 ms.
Figure 22: Evolution of calculation for a corner located dummy at 10 m/s speed crash. Frames are every 1 ms.

7. Discussion

Once the charts have been analyzed, it is possible to conclude that injuries might occur to passenger users. Those possible injuries to passengers are mainly located on lower extremities.

A detailed analysis reveals that accelerations at the head are quite low, and therefore the HIC36/15 index shows values much lower than the value of 600, so no damage would be inflicted to the brain.

The neck index is above the lower safety limit for the axial tensile neck loading. This problem occurs for all dummy locations and velocities. Values are different for every load case, but at any rate, neck problems could arise on crashing against buffers.

If axial compressive neck loading is checked, there are damages just when the dummy is in a center location. The other locations show values which are very close to the lower limit, but figures are near enough.

Regarding fore/aft shear force, free fall from ground and 3rd floor is not enough to cause damages/injuries. From the 7th floor, those forces are above the limits for the center load case and below the limits for the others. In brief, the neck would be particularly injured on the center location and in the case of a 7th floor free fall. Damage should affect ligaments and vertebrae disks mainly due to axial tensile neck loading at the center of the car. Lateral and corner locations are much safer in case of a buffer crash.

There is no significant thorax damage for any load case. There are neither high decelerations nor great displacements for the thorax zone. Thoracic spine acceleration and Sternum deflection are not significantly high. Values are higher for the center location load case but below safety limits.

The hip area does not fall under safety limits except for the 7th floor free fall at the corner location.

Axial Compressive right femur force is above the safety limits in all circumstances. Figures 16 and 17 show values above limits for any location and load case. Even for a much limited velocity, injuries could occur on femur. In case of a 7th floor free fall, damage at center location could be quite severe. Lateral and corner positions are less safe for free falls at lower floors and safer at top floors, when compared to the center location. That is due to the deflection of floor and stiffness of the car.

There is no damage above limits for the medial and lateral tibia left plateau compressive forces, except when velocity is related to top floors free fall. In such case, center location of passengers is worse.

There are injuries at ankles at almost any load case and speed. Figures 19 and 20 show damage levels for left and right ankle. The position adopted determines level of damage for each ankle.

As a summary, it can be concluded that no lethal damage or injuries should occur on an elevator car buffer crash. Velocity levels are high enough to cover high-speed elevators which are also related to polyurethane buffers. While no fatal injuries should occur, different levels of damage could arise regarding speed and location from this crash. This is especially certain for the lower extremities, with the femur and ankles as the most critical areas. Critical areas such as head and thorax are quite safe, although the neck might suffer from problems in the disks and ligaments.

8. Concluding Remarks, Future Work, and Considerations

In general terms, protection against the highly unlikely case of overspeed governor and safety gear failure is good.

Bearing in mind that the studied velocities are quite high, protection is sufficient enough as far as the 7th floor. Below that floor with maximum velocity of 2.5 m/s, damage and injuries sustained are not really serious or fatal.

Above that 7th floor, injuries could be really severe for both femur and neck.

Our main conclusion is that protection for high speed elevators is good enough in case of buffer crash.

It must be pointed that the analysis and reseach is for a elevator with only one passenger but the results if there were more passengers should be quite similar. Likewise, the dummy of choice corresponds to the 50th percentile male. It must be mentioned that results might differ in the case of heavier people and that could be the case for kid dummies. The analysis has been done at high speed, and the latest polyurethane generation buffers, steel spring buffers, and lower velocities would lead to different results.

The possibility of a combined overspeed governor malfunction or safety gear failure is extremely low, bearing in mind that the elevator brake should as well fail to stop the elevator. Despite these extremely low possibilities, in May 12, 2009, six people were injured at London’s Tower Bridge when an elevator plunged 10 feet (3.05 meters) to the ground. Six people were taken to the hospital. Their injuries included broken bones; a husband and wife suffered leg injuries, while three others suffered minor leg injuries. Ten people were able to walk away unharmed.


The publication of this research paper has been possible thanks to the funding given by the Industry and Innovation Department of the Government of Aragon as well as by the European Social Funds, the Research Groups GEDiX and VEHIVIAL, according to Regulation (CE) no. 1828/2006 of the 8th of December Commission.


  1. F. Fiorino, “Tragedies in parallel—Colgan crew's indication of runaway trim once again points investigators to Beechcraft 1900 elevator system for crash clues,” Aviation Week & Space Technology, vol. 159, no. 9, pp. 42–42.
  2. A. Lozzi and P. Briozzo, “Failure of an inclined elevator,” Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering, vol. 214, no. 2, pp. 323–333, 2000. View at Scopus
  3. A. Miravete and E. Larrode, “El libro del transporte vertical,” in Grupo I+D Transportes Y VehícuLos. Área de Ingeniería e Infraestructura de los Transportes, University of Zaragoza, (Spanish).
  4. Livermore Software, “Ls-DynaKeword user’s manual,” Version 971, 2006.
  5. H. Mertz, “Biofidellity of the hybrid III head,” in Hybrid III: the First Human-Like Crash Test Dummy, SAE International, 1994.
  6. C. Kroell, D. Schneider, and A. Nahum, “Impact tolerance and response of the human thorax I and II,” in Hybrid III: the First Human-Like Crash Test Dummy, SAE International, 1994.
  7. C. Culver, R. Neatherly, and H. Mertz, “Mechanical necks with humanlike response,” in Hybrid III, the First Human-Like Crash Test Dummy, SAE International, 1994.
  8. R. Neatherly, “Analysis of chest impact response data and scaled performance recommendations,” SAE International, 1994.
  9. J. Horsch and L. Patrick, “Cadaver and dummy knee impact response,” in Hybrid III: the First Human-Like Crash Test Dummy, SAE International, 1994.
  10. P. Begeman and P. Prasad, “Human anckle impact response in dorsiflexion,” in Hybrid III: the First Human-Like Crash Test Dummy, SAE International, 1994.
  11. S. H. Backaitis, Hybrid III: the First Human-Like Crash Test Dummy, Edited by H. J. Mertz, SAE International, 1994.