EAS 4710

1
EAS 4710 Aerospace Design 2
8. Risk, Reliability and Safety
2
Reliability and safety

• Reliability is the probability of mission success
• Mission success has two aspects
• Probability of the safety of the crew
• Probability that mission objectives are met

3
Top-level functional failures that could lead to
loss-of-vehicle (LOV)

• Propulsion failure Engine malfunction
• Vehicle configuration failure Wing separation
• Containment failure Escape of gas or debris
• Vehicle environment failure Loss of ECLSS
• Externally initiated failure Lightning strike

4
X-15 propulsion failure
5
Soyuz-1 configuration failure
6
Apollo 1 ECLSS failure (fire)
7
Apollo 13 containment failure
Service module structure failure
LM and Apollo capsule
8
Challenger SRBM failure
Containment failure due to engine malfunction
STS-51 t73.2 seconds
9
Columbia configuration failure
Launch
reentry
Z200,000ft V13,000mph
Foam debris
10
Risk

• Three elements determine risk
• Initiating causes
• Hazardous condition
• Consequences

Risk Magnitude (Likelihood)X(Impact)
Risk
Probability
Consequences
11
Probabilistic Risk Assessment Modeling Mechanics
Functional event sequence Events leading to
consequences
Master logic diagram Hierarchy of initiating
events
Fault tree Failure possibilities tracked
Event tree Outcomes of events detailed
12
Risk Management

• Engineering control design out the risk (90
  effectiveness)
• Administrative control impose control procedures
  (50 effectiveness)
• Personnel control provide training methods (30
  effectiveness)

13
Risk Management Programs

• Program Definition define consequences and
  acceptable risk level
• Hazards Evaluation define scenarios, estimate
  likelihood and impact, thus risk
• Risk Reduction identify corrective actions to
  reduce risk levels
• Implementation verify actions, continue life
  cycle assessments, periodic review

14
Space Missions
STS-115, September 9, 2006
15
Distribution of LOV risk
Orbiter 39
SSME 37
Landing 5
ET 3
ISRB 16
16
Mission Phases for Two-stage Rocket Round Trip to
ISS
17
System Safety and Reliability

• Xi mission success during phase i
• xi mission is safe during phase i
• Rms probability of mission success
• Rcs probability of crew safety

18
Mission Depends on Serial Success

• Assuming serial success of all the mission phases
  1lt i lt n, then the probabilities become
• Rms Pr(X1X2.Xn)
• Rc Pr(x1x2xn)

19
Mission Phases are Independent

• Assume each of the mission phases are
  independent, in which case
• Rms Pr(X1)Pr(X2)Pr(Xn) (success)
• Rcs Pr(x1)Pr(x2)Pr(xn) (safety)

20
Mission Phases for Two-stage Rocket Round Trip to
ISS
21
Success of Mission Phases Depends on a Number of
Systems
22
Assume Systems are Independent and Operate
Successfully and Safely
For phase 1, the first stage ascent Pr(X1)
Pr(Y1) Pr(Y2) Pr(Y3) Pr(Y4) Pr(Y5) Pr(Y1)
Pr(y1) Pr(y2) Pr(y3) Pr(y4) Pr(y5) Yi success
of system i yi safe performance of system i
23
Powered and Unpowered Phases
Unpowered phases
Rms Pr(X4) Pr(X6) Pr(X7) Pr(X8)
Pr(X1) Pr(X2) Pr(X3) Pr(X5) Rms
R1R2
Powered phases
Broadest differentiation of mission phases
24
Bracket Reliability (1) Low Propulsion System
Reliability
For low propulsion reliability set R11 and set
probabilities for powered phases equal Pr(X1)
Pr(X2) Pr(X3) Pr(X5)Plow Rms (Plow)4
Then Plow (Rms)1/4
25
Mission Phases for Two-stage Rocket Round Trip to
ISS
26
Bracket Reliability (2) Equal Propulsion System
Reliability
For propulsion reliability about the same as the
other systems and with all phases of the mission
about equally reliable Rms (Psame)8
Then Psame (Rms)1/ 8
27
Relationship Between Mission Reliability and
Phase Reliability
28
Reliability Estimation
Best reliability estimates come from experimental
data. For a constant failure rate, estimate l by
the ratio of failures to total operating hours
maximum likelihood estimate (MLE). Or else the
ratio of r successful components to n total
components tested
Success parameter
Success parameter when rn
29
Reliability of past manned space missions
Man-rated engines
30
Apportionment goals success limited by powered
phases
Bracket phase reliability by Plow and
Psame Average R for Apollo and STS R0.963 Set
mission goal to, say, R0.96 and thus powered
propulsion reliability is 0.9898ltRlt0.9949
PlowltRltPsame Propulsion limiting
P(X1)P(Y1)Plow0.9898
31
Apportionment goal success of all phases equally
probable
All 5 phases equal P(X1)P(Y1)50.9898 or
P(Y1)0.9980 Then 0.9898ltRlt0.9980 is
required Averaged engine data R0.94 (omitting
Atlas, R0.9731) More detailed analysis with
better experimental data needed
32
Generic accident scenario for PRA
Benign end state
Undesirable end state
consequences
33
Probabilistic Risk Assessment Modeling Mechanics
Functional event sequence Events leading to
consequences
Master logic diagram Hierarchy of initiating
events
Fault tree Failure possibilities tracked
Event tree Outcomes of events detailed
34
PRA Modeling Mechanics
35
Master logic diagram
Damage event
Possible causes
Functional failures
36
Master logic diagram (continued)
Functional failures
Subsystem failures
37
Master logic diagram (concluded)
Component failures
Causes of failure
38
Orbiter aft compartment
39
SSME schematic
Oxidizer Pre-burner Valve (OPBV)
40
SSME Powerhead
41
Functional event sequence for given failure event
S/Dengine shutdown
41
Space Access Vehicle Design
42
Event tree for a functional event
Undetected
43
Fault tree for a given failure
Space Access Vehicle Design
43
44
The Weibull random variable PDF
g a shape factor tc a characteristic time
45
The Weibull distribution
46
Cumulative distribution function
Area Probability of failure before time t
47
System reliabiity
Area System reliability
48
Weibull reliability and failure rate
49
Failure rate and shape factor
50
Failure rate and shape factor

• A constant failure rate is typical of system
  behavior at times after being broken-in but
  before being worn-out
• For shape factors glt1 the failure rate is
  inversely proportional to time. This is typical
  of early times, before the system is broken-in,
  when faulty components are weeded out.
• Conversely, for shape factors ggt1 the failure
  rate increases with time, as the system wears
  out.

51
Weibull expected time to failure
(Mean time to failure)
52
Failure models reliability estimate
Assuming g1 the failure rate z(t) 1/tc
(suitable for a 1st-order approximate analysis)
z(t) 1/tc R(t) e-t/tc MTTF
tc
tc,i constant failure rate for phase i titime
from beginning of phase i
53
Weibull distribution from test data
ag b -glntc
y
x
yaxj b