Transcript Reliability Engineering for Medical Devices

Reliability Engineering
Richard C. Fries, PE, CRE
Corporate Manager, Reliability
Engineering
Baxter Healthcare
Round Lake, Illinois
Definition of Reliability
The probability, at a desired confidence level,
that a device will perform a specified function,
without failure,
under stated conditions,
for a specified period of time
Customer’s Definition of
Reliability
A reliable product:
One that does what the customer wants,
when the customer wants to do it
Reliability Basics
Reliability cannot be tested into a product
It must be designed and manufactured into it
Testing only indicates how much reliability
is in the product
Purpose of the Reliability
Group
Determine the weaknesses in a design
AND correct them
before the device goes to the field
Areas Covered by
Reliability
Electrical
Mechanical
Software
System
Failure Rate
Electrical Reliability
Time
Failure Rate
Mechanical Reliability
Time
Failure Rate
X-Axis
Theoretical Software
Reliability
Time
X-Axis
Failure Rate
X-Axis
Practical Software
Reliability
Time
X-Axis
Failure Rate
System Reliability
Time
Set the Reliability Goal
Based on similar equipment
Used as the basis for a reliability budget
Listed as Mean Time Between Failures
(MTBF) in hours or cycles
MTBF = the time at which 63% of the
units in the field will have failed
Minimum goal is ten years with a 98%
reliability
Parts Count Prediction
Uses MIL-HDBK-217
Indicates whether the design
approximates the reliability goal
Indicates those areas of the design with
high failure rates
Chemical Compatibility
Test plastics with typically used chemical
agents (alcohol, anesthetic agents,
cleaning agents)
Cleaning agents are the worst
Force Puller
Component Testing
Cycle/life testing of individual components
Comparison of multiple vendors of
components
Determine applicability for the intended
use
Philosophy of Testing
Test to have the units pass
Test with the addition of stresses to check
the margins of functionality
Types of Tests
Time terminated, failed parts replaced
Time terminated, no replacement
Failure terminated, failed parts replaced
Failure terminated, no replacement
Test until first failure
Test until all samples fail
Determining Sample Size
Uses Chi-Square table
SS = Chi-square Value(MTBF goal)/2
Chi-square value includes confidence level and
degrees of freedom = 2f+2
Component testing – 90% confidence level
Life testing – 95% confidence level
Sample Calculation
Want to test valves to be used for
2,000,000 cycles per year with a 10%
failure rate after 10 years
Reliability = e(-t/MTBF)
MTBF = -t/ln Reliability
= -20,000,000/ln 0.90
= 389,914,514 cycles
Sample Calculation
MTBF = 389,914,514 cycles
Number of Samples
10
50
100
Number of Cyles
89,777,817
17,955,563
8,977,782
Component Test Setup
Component Test Setup
Component Test Setup
Calculating Sample MTBF
MTBF = (# of samples)(length of test)
# of failures
Calculating MTBF Where No
Failures Occur
A sample MTBF cannot be calculated
A lower one-sided confidence limit is calculated
and the MTBF stated to be greater than that
number
One-sided limit = 2(#units)(test time)
Chi square value for the
confidence limit and 2
degrees of freedom
Sample Calculation for a No
Failure Test
10 valves are tested for 10,000 cycles
with no failures. Calculate using a 90%
confidence level.
One-sided limit = 2(10)(10,000)
4.605
= 43,431 cycles
MTBF > 43,431 cycles
HALT
Acronym for Highly Accelerated Life
Testing
Used to find the weak links in the design
and fabrication process
Usually performed during the design
phase
HALT Testing
Possible stresses that can be applied:
random vibration
rapid temperature transitions
voltage margining
frequency margining
The product is stressed far beyond its
specifications
The test can be set up to find the destruct
limits
HALT Chamber
Goal of HALT Testing
Overstress the product
Quickly induce failures
By applying the stresses in a controlled,
stepped fashion, while continuing
monitoring for failures, the testing results
in the exposure of the weakest points in
the design
This test, if successful, will expose weak
points in the design
Environmental Testing
Operating temperature/humidity
Storage temperature/humidity
EMC
Surges/transients
Brown-outs
Electrocautery
Cell phones
ESD
Altitude
Environmental Testing
Autoclave
Shock
Vibration
Shipping
Tip testing
Threshold testing
Temperature Chamber
Walk-In Temperature
Chamber
Autoclave Testing
Customer Misuse
Excess weight on tabletop
Fluid spillage
Cross connection of wires
Pulling unit by non-pulling parts
Wrong order of pressing keys
“Knowing” how to operate the unit without
reading the manual
Making a Design Foolproof
The biggest mistake engineers make
when trying to make a design
completely foolproof
is underestimating the ingenuity
of complete fools
Failure Analysis
Failure: device does not operate according to
its specification
Determine root cause of the failure
Suggest methods to address the failure
Prototype Front Panel
Plastic Structure
Plastic Structure
Autoclave Testing
Manifold Port
Prototype Port
Life Testing
Operate the device in its typical
environment and application
Use appropriate on/off cycles
Can be used to verify the reliability goal or
a specific period of time, such as the
warranty period
Tracking Reliability Growth
in the Field
Collect manufacturing data on how many
units were manufactured by month
Collect field failure data, by month
Develop a reliability growth chart
Reliability Growth Example
Ventilator Reliability Growth
MTBF (Hours)
80000
60000
40000
20000
0
1997
1998
1999
Year of Report
2000
Reliability Growth Example
MTBF (hours)
Ventilator Reliability Growth
50000
0
1996
1997
1998
1999
Year of Report
2000
Reliability Growth Example
200000
Pre-June, 1997
Build
150000
100000
Post-June, 1997
Build
50000
0
19
97
19
98
19
99
20
00
20
01
20
02
20
03
MTBF (hours)
Estimate of Two Vaporizer Builds
Year of Build
The Reliability Group
You make it,
We’ll break it