The CorAide centrifugal blood pump developed at The Cleveland Clinic Foundation consists of three subassemblies: a volute housing, a stator assembly, and a rotating assembly. An inverted motor distinguishes the device from other blood pumps, while the noncontacting hydrodynamic, blood lubricated journal bearing makes it nonwearing. The pump has only one moving element that rotates, thus propelling blood through the device. 1,2 The purpose of this study was to assess the effects of impact on the hydraulic, vibration, and magnetic performance of the ventricular assist device in situations likely to take place in real life occurrences, such as slipping and falling while bathing or on stairs.
Methods and Materials
The first test involved dropping the pump assembly onto a hardwood surface from a height of 1 m (3.25 ft) under nonoperational conditions to simulate an accidental drop of the pump assembly during packaging or handling. The remaining two tests were performed from 1.2 and 4 m (4 and 13 ft), with the pump operating in a mock circulatory loop containing blood analog solution, at nominal operating conditions (pump flow of 5 L/min, pressure change of 100 mm Hg, at 2,750 rpm). Arranging the instrumented pump and the mock loop in a box surrounded by rigid dry foam controlled the orientation of the pump. The boxed pump and mock circulatory loop were dropped onto padding in three different orientations, causing loading in the radial direction, axial direction toward the secondary impeller, and axial direction toward the primary impeller (Figure 1). Padding thickness was adjusted and the height was selected so that the total impact duration (deceleration time) and peak acceleration were comparable with chest accelerations from published automotive test data for survivable crashes. 3 During impact testing, the pump assembly was instrumented with a triaxial accelerometer having a ± 50 g range with an output of 100 mV/ g (PCB Piezotronics, Model # 356B08, Depew, NY). The signals for the X, Y, and Z axes were acquired using PowerLab software (AD Instruments Inc., Mountain View, CA).
Visual inspection of all blood contacting surfaces was performed before and after each series of drop tests to document any mechanical defects incurred as a result of impact. Any defects found before testing were documented and tracked throughout the testing process.
Hydraulic Performance Evaluation
The CorAide device was connected to a continuous flow mock circulatory loop containing blood analog solution that consisted of 45% glycerin solution in deionized water, achieving a viscosity of 3.5 ± 0.20 cP at 37 ± 2°C. An ultrasonic flow probe (Transonic Systems Inc., Ithaca, NY) measured pump flow, while inlet and outlet pressures were monitored via pressure transducers located within 6 inches of the pump ports. Testing was performed at speeds of 2,500, 2,750, 3,000, and 3,200 rpm and flow was varied to produce 5 to 10 points between zero and maximum flow, distributed in increments equal within 20%. Flow resistance was manipulated with an adjustable clamp on the pump outflow tubing; loosening the clamp increased pump flow. Speed (rpm), pump flow in L/min (Q), inlet pressure in mm Hg (Pin), and outlet pressure in mm Hg (Pout) were recorded for hydraulic performance evaluation. From the collected data, hydraulic power and pump efficiency were calculated usingMATHMATH
The conversion factor of 0.00222 was used to obtain units of hydraulic power in watts. To evaluate the hydraulic performance, electric power consumption and efficiency were compared at nominal operating conditions before and after each series of impact tests were performed. The nominal operating conditions are conditions under which the CorAide pump was designed to operate; pump flow is 5 L/min at a pressure change of 100 mm Hg across the pump at approximately 2,750 rpm.
Vibration Performance Evaluation
Vibration and hydraulic performance testing were performed simultaneously, immediately before and after each series of drop tests. A uniaxial accelerometer having a ± 500 g range with an output of 10 mV/ g (PCB Piezotronics, Model # 303A02, Depew, NY) was used to evaluate the vibration of the pump in the radial direction. The signal from the accelerometer was filtered using a bandpass filter with a bandwidth of 10 to 10,000 Hz at a gain of 40 dB. The amplitude of vibration at the first harmonic frequency (fundamental frequency) for the nominal operating condition before and after each set of impact tests was compared to evaluate any changes in vibration performance that may have resulted from impact.
Magnetic Induction Evaluation
Magnetic induction evaluation was performed on the rotating assembly before and after all drop testing was conducted. A special fixture was designed to secure the rotating assembly in place on a rotary table equipped with a degree indicator. In addition, a probe holder was designed to hold the Hall probe (F.W. Bell, Model #STH57–0,404, Orlando, FL) securely at three different depths within the rotating assembly at the top, middle, and bottom. The Hall probe and a gaussmeter (F.W. Bell, Model #5,070, Orlando, FL) were used to measure the inductance of the rotating assembly every 10° throughout its entire circumference. The inductance as a function of rotation angle was plotted and comparison was made before and after the drop testing.
The accelerometer signals during the drop onto a hardwood surface from 1 m (3.25 ft) under nonoperational conditions were clipped because the amplitudes of acceleration were out of the range of the accelerometer. The horizontal, linear behavior of the X and Y acceleration signal in Figure 2 depicts the portions of the signals that were out of range. Therefore, actual peak acceleration of the pump could not be quantified.
Impact testing of the operating pump from a height of 1.2 m (4 ft) in the three orientations described previously resulted in peak resultant accelerations ranging from 12.48 to 15.5 g, with total deceleration time ranging from 115 to 120 ms (Figure 3). Impact testing from a height of 4 m (13 ft) resulted in peak resultant accelerations ranging from 34.49 to 44.01 g, with total deceleration time ranging from 104 to 120 ms (Figure 4).
Hydraulic Performance, Vibration, and Magnetic Induction
With hydraulic and vibration evaluation before and after impact, it was seen that there was no significant change in pump performance, as demonstrated by the calculated power consumptions of 6.35 and 6.12 W and efficiencies of 16.5 and 17.0% before and after impact testing at nominal operating conditions. In addition, there was no significant change in the vibration traces captured during performance testing before and after the drop. Visual inspection before and after impact showed that there was no mechanical damage incurred by any pump component.
Magnetic induction testing was performed before and after all impact testing. The overall inductance and residual magnetic imbalance of the rotating assembly were 3.10 k Gauss and 6.9%, respectively, before impact testing. After the completion of impact testing, the overall inductance and residual magnetic imbalance were measured as 3.09 k Gauss and 4.9%, respectively, suggesting that the magnetic properties of the rotating assembly did not change as a result of impact testing.
The testing configuration used in this experiment was chosen in an attempt to simulate situations that the CorAide device might be exposed to either in manufacturing or clinical use. The first test addressed manufacturing issues and involved dropping the instrumented pump onto a hardwood surface under dry, nonoperational conditions. It was intended to determine if an accidentally dropped pump is still usable. X and Y acceleration signals were clipped at 10 and 34 g, respectively, rather than the upper limit of the accelerometer range, 50 g, as one might expect. This is due to the fact that the next data points for the X and Y signals exceeded the 50 g limit and, therefore, saturation of the signals occurred. Although peak accelerations were not obtained, it can be concluded from pre- and postimpact pump evaluation that impact from the height of 1 m (3.25 ft) had no effect on pump performance or durability.
All vibration traces contained two main peak accelerations. The first was at a frequency of approximately 45 Hz, which is the fundamental frequency of the rotating assembly (once per revolution) at a pump speed of 2,750 rpm. Increases in its amplitude would suggest changes in the magnetic balance of the rotating assembly, as would the appearance of additional accelerations. The second peak acceleration at 550 Hz is a result of the motor design. The stator windings within the stator housing have 12 slots around its circumference. Magnetic alignment with each slot occurs 12 times in each revolution of the rotating assembly. The frequency of this harmonic is 12 times that of the fundamental frequency of the pump and, therefore, will always be present. Typical magnitudes are in the range of 0.2 to 0.4 g, resulting in a displacement of only 10 microinches at 550 Hz, which is not of tangible concern.
Tests conducted while the pump was operating were intended to simulate situations that may take place during clinical use. Unfortunately, no data were found for acceleration of the thorax during normal, everyday activities such as jogging, descending a flight of stairs, or slipping and falling in the bathtub. Data regarding human exposure to shock established the limit of voluntary human exposure as 10 to 15 g for < 100 ms in headward accelerations, where no injury or debilitation resulted. 4 To reproduce pump accelerations in this range, a height of 4 ft was chosen for the second set of impact tests. The results from this test produced pump accelerations of 12.48 to 15.50 g for 115 to 120 ms.
Extreme conditions were examined by researching voluntary human exposure limits that did not produce life threatening injury, where the upper limit of voluntary human exposure was determined as 40 to 45 g for a duration of approximately 100 ms for spineward acceleration. 4 However, this threshold is not related to any type of daily life activities. On the other hand, the automotive industry has collected extensive data while developing the Hybrid III anthropomorphic crash test dummy. Cadaver chest accelerations in 30 mph frontal crash simulations using three-point constraints were used to determine the desired input for impact testing of peak acceleration and impact duration. 3 The impact event for a 30 mph frontal crash lasted for a duration of approximately 100 to 120 ms, with peak accelerations on the order of 20 to 40 g.
In the automotive industry, injuries are described by Abbreviated Injury Scale (AIS) ratings for severity. 5 The AIS ratings range from 0 = no injury, to 6 = maximum injury (currently untreatable). Cadaver studies used in 30 mph frontal impact simulation showed that only 19% of occupants sustain injuries having an AIS rating > 1, 3 and the probability of survival (AIS ≤ 3) has been shown to be at least 0.84. 6 Although a patient may never be in this type of collision, the device should still be capable of withstanding any normally survivable accident experienced by a patient. Therefore, it was decided that this situation would be tested. Several iterations were performed to obtain the desired peak accelerations and impact durations. This test did not mimic the in vivo response of the pump to impact, nor did it simulate the effects of attenuation that the tissue response would have contributed. The objective of this testing was to impart similar accelerations as seen in the thoracic environment of cadavers used in automotive testing. This test produced peak resultant pump accelerations of 34.49 to 44.01 g for 104 to 120 ms. As seen by these results, the peak accelerations measured during these tests slightly exceeded the desired range.
Any physical damage that might have been sustained to the pump components would have been observed during visual inspections that were performed before and after each series of impact tests. Damage to the stator windings would have been evidenced by significant differences in power consumption before and after each series of impact tests. Furthermore, magnetic imbalance of the magnet ring within the rotating assembly would have resulted if the magnet ring were cracked or damaged during impact. Evidence of magnet damage would have been manifested in the vibration traces, causing alterations in these traces either by the change in amplitude of vibration at a given frequency or the appearance of additional amplitudes of acceleration at other frequencies not shown in the traces before testing. There would also have been marked differences in the preimpact testing versus postimpact testing induction traces.
The intent of this experiment was to examine the effects of the CorAide assist device under impact conditions that simulate realistic situations, such as a 1 m drop onto a hard surface, and operating during and after an accidental fall. This intent was fulfilled by the three series of tests that were performed. The resulting peak accelerations and impact durations were within the desired ranges specified. Furthermore, there was no obvious effect on hydraulic performance, vibration, durability, or magnetic properties. These tests show that this pump will perform as designed during a myriad of potential events that a patient may encounter in normal, everyday life.
1. Horvath D, Golding L, Massiello A, et al: The CorAide Blood Pump. Ann Thorac Surg 71: S191, 2001.
2. Ochiai Y, Golding L, Massiello A, et al:In vivo
hemodynamic performance of the Cleveland Clinic CorAide blood pump in calves. Ann Thorac Surg 72: 747–53, 2001.
3. Saul R, Sullivan L, Marcus J, Morgan R: Comparison of current anthropomorphic test devices in a three-point belt restraint system, in Backaitis SH, Mertz HJ (ed), Hybrid III: The First Human-Like Crash Test Dummy
. Warrendale Pennsylvania: Society of Automotive Engineers, 1994, pp 603–620.
4. von Gierke HE, Brammer AJ: Effects of shock and vibration on humans, in Harris CM (ed), Shock and Vibration Handbook
, 4th ed. New York: McGraw-Hill, 1996, pp 44.1–44.67.
5. Benson B, Perl T, Smith G: Lateral load sensing hybrid III head, in Backaitis SH, Mertz HJ (ed), Hybrid III: The First Human-Like Crash Test Dummy
. Warrendale Pennsylvania: Society of Automotive Engineers, 1994, pp 603–620.
6. Federal motor vehicle safety standards: Side impact protection. Fed Regist 55: 722–745, 1990.