Gait Evaluation of a Trans-Femoral Prosthetic Simulator

A device was developed to allow non-amputees to walk like a person with a trans-femoral (TF) amputation. This prosthetic simulator was validated by fitting, training, and testing five subjects in a motion analysis laboratory. Sagittal plane kinematic and kinetic analyses showed that joint mechanics during walking were similar between the test subjects and comparative results from the literature. The test subjects walked slower and move their hip and knee joints faster (higher angular velocity values during terminal swing) than results from the literature, although this result was not statistically significant (p<0.05). These findings were consistent with new prosthetic users who are more tentative during gait training; however, a perfect simulation would show no difference in kinematic results. These results support the use of a TF prosthetic simulator to help health care providers experience the prosthetic fitting process from a client's perspective..


Prosthetic gait has been well documented over the last few decades. By reviewing research papers and textbooks, health care providers can understand the theory behind (TF) amputee locomotion (1-12). However, the written word has difficulty conveying the physical experience of walking with a prosthesis. Health care practitioners would benefit by experiencing the exertion, balance requirements, and insecurities experienced by people who must learn to walk with a prosthesis.

To provide a real-life prosthetic experience for non-amputees, Nielen (13) developed a prosthetic/orthotic simulator that allows a non-amputee to walk as an TF amputee (fig 1). This device consisted of a prosthetic foot and pyramid pylon, two uprights attached to orthotic knee joints, an elastic extension assist, a lower-leg sling, and TF prosthetic socket with the distal portion removed. The extension assist consists of two nylon straps that connect to a proximal uprights and to an elastic strap (the elastic strap connects to the distal end of the pylon).

Once a socket size was selected and the simulator was bench aligned, the user stepped through the socket, flexed their knee, and secured their lower leg in position with the sling attachment. The typical user progressed from walking between parallel bars, to walking with a cane, and finally to independent ambulation. The students and staff who used the prosthetic simulator considered the experience valuable and empathy-building.

Figure 1:


Five able bodied subjects, three female and two male, were recruited from The Rehabilitation Centre and the University of Ottawa. These subjects averaged 27.6 (standard deviation[SD] = 6.11) years, 1.62 (SD = 0.07) m, and 70.0 (SD = 13.5) kg. These subjects had no previous experience with the prosthetic simulator and had no physical limitations that would affect their ability to ambulate with the device.

The prosthetic simulator used in this study was similar to the device reported by Nielen(13). Based on the IPOSa measurement procedures, the project prosthetist used the IPOS computer-aided design/computer-aided manufacture (CAD/CAM) ischial containment brim library to generate a series of unmodified socket shapes that accommodated the subject's anatomical dimensions. For each subject, the prosthetist attached the appropriate trimmed socket to the simulator uprights and bench aligned the device using the subject's shoe. While wearing Lycra tights and standing between parallel bars, the subject stepped through the socket and started weight-bearing on the simulator. The prosthetist addressed initial socket fit problems before the subject's leg was secured in flexion.

Once the simulator has been properly configured, each subject completed two gait training sessions. A physiotherapist from the Rehabilitation Centre's amputee rehabilitation team was available to help with the training process. After completing the 45 to 60 minute sessions, all subjects were able to walk short distances unassisted. All subjects felt more secure with one cane when walking over longer distances.

Following gait training, data were collected from each subject in the Rehabilitation Centre's Gait and Motion Analysis Laboratory. All subjects wore a dark shirt, dark Lycra tights, and athletic shoes during data collection. After donning the prosthetic simulator, reflective markers were attached to the subject's shoulder, hip, knee, ankle, heel, ball, and toe. The project's prosthetist reviewed the socket fit and simulator setup before starting the data collection process.

While using a cane, the subject performed a series of warm-up trials along a 10m walkway. During these warm-up trials the subject's start point was positioned such that, while walking at their natural cadence, they contacted an AMTIb force plate.

Six data collection trials were completed for each subject as they walked along the walkway - three trials with a cane and three trials without a cane. The motion was video-taped at the same time as force plate data were collected. The Ariel Performance Analysis Systemc (APAS) was used to digitize the reflective markers (60Hz) and collect the force plate data (200 Hz) for input into the BIOMECH analysis packaged. BIOMECH combined the marker and force plate data to calculate kinematics and kinetics at the ankle, knee, and hip. Heel strike was used define the starting point of each stride. Results from the three trials were ensemble averaged to produce representative data for each subject.

T-tests (p < 0.05) and descriptive statistics were used to statistically compare stride parameter data between the literature (1-11,14) and the cane trials and between the literature and the no-cane trials. Descriptive statistics and were used to compare kinematic and kinetic analysis results with the literature. Kinematic and kinetic curves were also compared on the basis of shape and on the similarity to gait events established by Winter 12.


All subjects were able to progress from parallel bar walking to unassisted walking in two training sessions. At least 48 hours was required between training sessions to allow the subjects to recover from the initial training. The subjects reported muscle stiffness, fatigue, and occasionally chaffing from their training sessions.

As shown in table 1, all subjects walked slower than experienced TF amputees.

Table 1: Average stride parameters (standard deviations in parentheses).

  Stride Length (m) Stride Time (s) Speed (m/s)
Cane 1.24 (0.10) 1.59 (0.12) 0.79 (0.10)
No-cane 1.26 (0.14) 1.56 (0.08) 0.82 (0.11)
Literature average (1-8) 1.33 (0.06) 1.38 (0.04) 0.99 (0.05)

A cane, while providing security, slowed down the subject's walking speed. Both decreased stride length and increased stride time contributed to this reduction in speed. The differences between simulation and averaged stride parameters from the literature were not statistically significant (p < 0.05). Stride parameter statistical powers were over 0.97 for all measures except stride length (0.74 with cane, 0.23 without cane)15.

The kinematic measurements were similar to data from the literature1-12. Ankle joint angular velocities were very low for all trials (mean: cane = 0.75 deg/s, no cane = 0.85 deg/s).

Kinetic analysis results were also similar to data from the literature. Ankle moments had the characteristic dorsiflexor moment after heel strike and plantar flexor moment towards toe-off (fig 2). All ankle powers were very low, under 0.43 W/kg. The low power and angular velocity values indicated limited motion at the simulator's semi-rigid foot/ankle unit and its inability to generate energy. Average ankle standard deviation values were 0.31 deg/s for angular velocity, 0.09 N.m/kg for moment, and 0.09 W/kg for power.

Figure 2:

Knee moment results were also similar to data from the literature12 (fig 3). A knee flexor moment occurred from heel strike until push-off, as the knee was maintained in extension during body weight support. During push-off and the initiation of swing, an eccentric extensor moment controlled knee flexion. In late swing, an eccentric flexor moment controlled knee extension in preparation of the heel strike. The low power values are expected due to the low shank segment mass. Knee joint angular velocity values were at or above angular velocity values for non-amputees. Average knee standard deviation values were 0.89 deg/s for angular velocity, 0.15 · N.m/kg for moment, and 0.20 W/kg for power.

Figure 3:

A concentric hip extensor moment is present during the first half of stance (fig 4). This moment acts to help bring the body up over the prosthetic limb. As the body pass midstance, an eccentric hip flexor moment acted to control hip extension. During pushoff, a strong concentric hip flexor moment was necessary to pull the leg upward and forward. This action also acted to flex the knee and initiate the swing phase of gait. The hip was relatively inactive during the middle of the swing phase. At the end of swing a highly variable, but phasic, pattern of concentric and eccentric extensor moments occurred as the subjects prepared the prosthesis for the next heel-strike (i.e., ground contact with the knee extended and the foot forward). Average hip standard deviation values were 0.60 deg/s for angular velocity, 0.22 N.· m/kg for moment, and 0.27 W/kg for power.

Figure 4:

As show in table 2, the mean work results were similar to the results of Seroussi, et al.14 for many phases in the gait cycle. Work results during the K3 phase, late stance and early swing, were approximately 40 percent lower with the simulator. This result may be attributed to the differences between the simulator's free moving knee joints and the stance/swing phase control knee joints used in Seroussi, et al.14. The Mauch SNS hydraulic joint is superior at providing eccentric extensor activity to control the knee. The differences in knee joint hardware could also account for the 54 % difference in H2 work values (i.e., the eccentric hip flexor activity during late stance). More work was done for the cane and no-cane trials to initiate swing (H3). The inexperienced simulator subjects produced more work at the hip to ensure that the foot cleared the ground during swing. The average duration of concentric hip flexor activity during the H3 phase was longer for the simulator subjects as compared to the above-knee amputation subjects from Seroussi, et al.14. Our hip work results during late-stance and early-swing were more similar in duration with the results from Cappozzo, and colleagues. 1.

Table 2: Mean positive and negative work (J) for all subjects based on the phases defined by Winter 12. Standard deviations are in parentheses.

  Cane No cane Seroussi, et al. (15)
A2 2.7(1.5) 2.7(1.7) 4.9(2.1)
K3 -9.3(2.2) -9.1(3.2) -15.3(8.8)
K4 -4.5(1.1) -5.1(1.8) -4.1(.07)
H1 7.8(5.7) 6.9(4.6) 6.0(5.7)
H2 -3.6(3.3) -6.3(5.2) -13.6(5.6)
H3 15.1(5.4) 14.5(4.5) 8.7(3.5)


In the field of physical rehabilitation, health care professionals can benefit by using the mobility aids they prescribe to their clients. Driving a wheelchair or walking with a knee-ankle-foot orthosis can help sensitize clinicians towards their client's situation. Setting up such a system for prosthetics is a more difficult task since typical prosthetic devices cannot be used on non-amputees. The results from this study have shown that Nielen's 13 prosthetic simulator design is an effective tool for allowing non-amputees to experience TF prosthetic gait. Initial feedback from the subjects in this study confirms that there is more to learn about prosthetic use than just balance and walking mechanics. All of the subjects were extremely fatigued after the training sessions. Many reported muscle soreness and chaffing from the socket as a result of the first few times on the simulator. TF prosthetic clients must deal with these factors during prosthetic fitting and early gait training.

Stride parameter results showed that, on average, the test subjects walked slower than experienced prosthetic users. The decrease in walking speed was based on a shorter stride length and a longer stride time. Inexperienced prosthetic users, such as the subjects in this study, shorten step length and lengthen their stride time to compensate for their lack of "prosthetic coordination" and confidence. These results are to be expected from clinicians who use the prosthetic simulator since they will not usually receive the same level of gait training as a new amputee.

The kinematic and kinetic analysis results were similar to gait analyses from people with TF amputations. Motion at the foot/ankle section was very small and highly variable between subjects. Much of this variability occurred between foot flat and toe-off. During this period, the subjects were likely trying to control the prosthesis during single support. While intersubject variability was higher than results reported in the literature for amputee gait, the averaged results at the ankle were consistent with the literature on TF ambulation(1-12).

The knee moment and power results were similar in shape and magnitude to the results from Winter 12. The angular velocity curve, while being of the expected shape, exhibited higher than expected velocities in late stance. One explanation for this result is that the simulator used free-moving knee joints as opposed to a dampened joint that is commonly used for TF prostheses. In addition to the lack of dampening in extension, the subjects were accelerating and then decelerating their hip during this period (i.e., flex the knee joint to clear the ground and then extended the leg to prepare for heel strike). While a smoother walking gait could be expected by reducing the hip and knee angular velocities, clinicians using the simulator should be advised to consider safety their primary objective (i.e., ensuring that the leg is fully extended at heel-strike by accelerating the knee joint during late swing). Except for the excessively high hip angular velocities during late stance, hip motion was similar to results from the literature.


This study showed that a simulator13 is capable of providing non-amputees with the experience of walking with a TF prosthesis. Ambulation with the prosthetic simulator produces stride parameter, joint kinematic, and net joint kinetic results that are consistent with results from the literature. This simulator should be of benefit to health care providers who work with people with amputations by sensitizing them to the requirements for prosthetic locomotion.


The authors would like to acknowledge the assistance of Alain Plouffe and the Gait and Motion Analysis Team at The Rehabilitation Centre (Ottawa) toward completion of this project.


  1. Cappozzo A, Figura F, Leo T, Marchetti, M. Biomechanical evaluation of above-knee prostheses. Biomechanics V-A - Proceedings of the Fifth International Congress of Biomechanics, ed. Paauo V. Komi, Baltimore, University Park Press, 1976: 366-372.
  2. Hale, SA. Analysis of the swing phase dynamics and muscular effort of the above-knee amputee for varying prosthetic shank loads. Prosthet Orthot Int 1990; 15:125-135.
  3. Hale SA. The effect of walking speed on the joint displacement patterns and forces and moments acting on the above-knee amputee prosthetic leg. J Prosthet Orthot 1991; 3(2): 59-78.
  4. James U, Öberg K. Prosthetic gait pattern in unilateral above-knee amputees. Scand J Rehabil Med 1973; 5: 35-50.
  5. Murray MP. Gait patterns of above-knee amputees using constant-friction knee components. Bull Prosthet Res 1980; 17(2); 35-45.
  6. Murray MP, Mollinger LA, Sepic SB, Gardner GM, Linfer MT. Gait patterns in above-knee amputee patients: hydraulic swing control vs constant-friction knee components. Arch Phys Med Rehabil 1983; 64: 339-345.
  7. Tashman S, Hicks R, Jendrzejczyk DJ. Evaluation of a prosthetic shank with variable inertial properties. Clin Prosthet Orthot 1985; 9: 23-28.
  8. Zuniga EN, Leavitt LA, Calvert JC, Canzoneri J, Peterson CR. Gait patterns in above-knee amputees. Arch Phys Med Rehabil 1972; 53: 373-382.
  9. Hershler C, Milner M. Angle-angle diagrams in above-knee amputee and cerebral palsy gait. Am J Phys Med 1980; 59(4): 165-183.
  10. Lewallen R, Quanbury AO, Ross K, Letts RM. A biomechanical study of normal and amputee gait. Biomechanics IX-A - Proceedings of the Ninth International Congress of Biomechanics, ed. D. Winter, Illinois, Human Kinetics, 1985: 587-592.
  11. Yang L, Solomonidis SE, Spence WD, Paul JP. The influence of limb alignment on the gait of above-knee amputees. J Biomech 1991; 24(11): 981-997.
  12. Winter DA. The Biomechanics and Motor Control of Human Gait: Normal, Elderly, and Pathological. Waterloo: Waterloo Biomechanics, 1991.
  13. Nielen D. Transfemoral prosthesis simulator - an educational tool. Alignment 1997: 66-67.
  14. Seroussi RE, Glitter A, Czerniecki JM, Weaver K. Mechanical Work Adaptations of Above-Knee Amputee Ambulation. Arch Phys Med Rehabil 1996; 77: 1209-1214.
  15. Dupont WD, Plummer WD. Power and Sample Size Calculations: A Review and Computer Program. Controlled Clinical Trials 1990; 11: 116-128.


  1. IPOS Orthopedics Industry, Zeppelinstrassse 30, 21337, Luneburg, Germany.
  2. AMTI, 176 Waltham St, Watertown, MA 02172.
  3. Ariel Dynamics, Inc., 4891 Ronson Ct, Se F, San Diego, CA 92111.
  4. University of Ottawa,Ottawa, Ontario, Canada K1H 8M2.


  1. Average stride parameters (standard deviations in parentheses).
  2. Mean positive and negative work (J) for all subjects based on the phases defined by Winter 12. Standard deviations are in parentheses.


  1. Therapist wearing the trans-femoral prosthesis simulator.
  2. Ankle angular velocity, moment, and power ensemble averages for all subjects.
  3. Knee angular velocity, moment, and power ensemble averages for all subjects.
  4. Hip angular velocity, moment, and power ensemble averages for all subjects.