Introduction to Reliability Theory

(part 2)

Frank Coolen

UTOPIAE Training School II, Durham University

3 July 2018

(UTOPIAE) Introduction to Reliability Theory 1 / 21

Outline

Statistical issues

Software reliability

Decision making

Survival signature

Imprecision

Resilience

(UTOPIAE) Introduction to Reliability Theory 2 / 21

Statistical Issues

Random failure time T > 0 with CDF F (t) = P(T ≤ t)

Survival Function S(t) = P(T > t) = 1 − F (t)
pdf f (t) = F 0 (t) = −S 0 (t)
Hazard Rate h(t) = f (t)/S(t)
Interpretation, for small δt: h(t)δt ≈ P(T ≤ t + δt | T > t)
Rt
Cumulative Hazard Function (CHF) H(t) = 0 h(x)dx
R t f (x)
H(t) = 0 S(x) dx = − ln S(t)
Rt
S(t) = exp{−H(t)} = exp{− 0 h(x)dx}

(UTOPIAE) Introduction to Reliability Theory 3 / 21

Most statistical methods for estimation of parameters in assumed
models (e.g. Weibull) use the Likelihood Function, either maximising it
or combined with a prior distribution in Bayesian analysis.
Reliability data are often affected by right-censoring: an item has not
failed during a particular period of time.
Let t1 , . . . , tn be observed failure times, and c1 , . . . , cm right-censored
observations. For a parametric model with parameter θ, the likelihood
function is
n
Y m
Y
L(θ|t1 , . . . , tn ; c1 , . . . , cm ) = f (tj |θ) S(ci |θ).
j=1 i=1

This assumes that the censoring mechanism is independent of the

failure process.

(UTOPIAE) Introduction to Reliability Theory 4 / 21

Regression models for reliability data

For regression models in reliability applications Weibull models are

often used, with the survival function, depending on a vector of
covariates x, given by
  ηx 
t
S(t; x) = exp − .
αx

Some simple forms are often used for the shape and scale parameters
as functions of x, e.g. the loglinear model for αx , specified via
ln αx = x T β, with β a vector of parameters, and similar models for ηx .
The statistical methodology is then pretty similar to general regression
methods (e.g. fitting by MLE or Bayesian methods) and implemented in
statistical software packages.

(UTOPIAE) Introduction to Reliability Theory 5 / 21

Failure processes
A system fails at certain times, there may be some actions during this
period which affect failure behaviour, e.g. minimal repairs to allow the
system to continue its function, or replacement of some components,
or other improvements of the system. Normally this should not include
full replacement of the system.
Let the random quantities T1 < T2 < T3 < . . . be the failure times of
the system, let Xi = Ti − Ti−1 (with T0 = 0).

(UTOPIAE) Introduction to Reliability Theory 6 / 21

Rate of OCcurrence Of Failure (ROCOF)
Let N(t) be the number of failures in the period (0, t], then the ROCOF
is:
d
v (t) = EN(t).
dt
Increasing (decreasing) ROCOF models a system that gets worse
(better) over time.
Note that the ROCOF is not the same as the hazard rate (the
definitions are clearly different!), although intuitively they are a bit
similar. If we consider a standard Poisson process, with iid times
between failures being exponentially distributed, then the ROCOF and
hazard rate happen to be identical.

(UTOPIAE) Introduction to Reliability Theory 7 / 21

Non-Homogeneous Poisson Process models (NHPP)
The crucial assumption in these models, as for standard Poisson
processes which are just a special case, is that the numbers of failures
in distinct intervals are independent if the process characteristics are
known.
A NHPP with ROCOF v (t) is easiest defined by the property that the
number of failures in interval (t1 , t2 ] is a Poisson distributed random
quantity, with mean
Z t2
m(t1 , t2 ) = v (t)dt.
t1

(UTOPIAE) Introduction to Reliability Theory 8 / 21

Suppose we have observed the system over time period [0, r ], and
have observed failures at times t1 < t2 < . . . < tn ≤ r , then the
likelihood function is
( n )  Z r 
Y
L= v (ti ) exp − v (t)dt .
i=1 0

This enables inference as usual.

Two basic, often used parametric ROCOFs are

v1 (t) = exp(β0 + β1 t)

and
v2 (t) = γηt η−1 ,

(UTOPIAE) Introduction to Reliability Theory 9 / 21

Software Reliability

Many models that have been suggested, during the last five decades,
for software reliability, are NHPPs which model the software testing
process as a fault counting process.
Many models are variations to a basic model which assumes:
(1) Software contains an unknown number of bugs, N.
(2) At each failure, one bug is detected and corrected.
(3) The ROCOF is proportional to the number of bugs present.

(UTOPIAE) Introduction to Reliability Theory 10 / 21

The basic model is a NHPP with

v (t) = (N − i + 1)λ, for t ∈ [Ti−1 , Ti ),

for some constant λ.

N and λ are both considered unknown, and estimated from data,
where N tends to be of most interest, or in particular the number of
remaining bugs.

(UTOPIAE) Introduction to Reliability Theory 11 / 21

Software reliability is a very important topic area, in particular
statistical support for software testing is challenging.
Book to appear (Wiley, August 2018): Analytic Methods in Systems
and Software Testing, edited by Keneth, Ruggeri and Faltin.
Includes chapter by Wooff, Goldstein and Coolen on the use of
Bayesian Graphical Models for software testing, including many
aspects of implementation.
Challenge: considering reliability of systems with hardware and
software, including interaction.

(UTOPIAE) Introduction to Reliability Theory 12 / 21

Decision Making
Traditionally, decisions related to reliability (e.g. planning of
replacements or inspections, warranties, etc) were mainly based on
Renewal (Reward) Theorem:
It is assumed that the same process goes on for a very long time
(‘infinitely’), consisting of consecutive cycli which are stochastic copies
of each other.
Let Xi be the random length of cyclus i, and Ri a random reward
associated with cyclus i. Assume that the Xi ’s are iid, and that the Ri ’s
are iid, but allow dependence of Ri on Xi .
For a process starting at time 0, let R(t) be the random cumulative
reward upto time t. Then

R(t) E(Ri )
→ if t → ∞.
t E(Xi )

So ‘long-run average reward per unit of time’ is used.

(UTOPIAE) Introduction to Reliability Theory 13 / 21
Assumption underlying use of renewal reward theorem often
unrealistic; optimisation over one cycle, or a few cycli, or a fixed period
of time more realistic. Analysis may be harder, and decisions typically
‘less risky’.
In recent years, emphasis has shifted to service contracts of fixed
length. This is a main game-changer, with possible advantages for
reliability theory (e.g. wrt data ownership) and many new challenges
for decision making, e.g. planning and location of spare parts.

(UTOPIAE) Introduction to Reliability Theory 14 / 21

Survival Signature

System with K ≥ 2 types of components

PK
mk components of type k ∈ {1, 2, . . . , K }, with k=1 mk =m

State vector x = (x 1 , x 2 , . . . , x K ), with x k = (x1k , x2k , . . . , xm

k ) the
k
sub-vector representing the states of the components of type k with
xik = 1 if the ith component of type k functions and xik = 0 if not.
Structure function φ(x) = 1 if system functions with state x and
φ(x) = 0 if not.

(UTOPIAE) Introduction to Reliability Theory 15 / 21

The Survival Signature Φ(l1 , l2 , . . . , lK ), with lk = 0, 1, . . . , mk , is the
probability that a system functions given that precisely lk of its
components of type k function, for each k ∈ {1, 2, . . . , K }.

There are mlkk state vectors x k with precisely lk of their mk

components xik = 1, so with m

P k k
i=1 xi = lk .

Let Sl1 P
,...,lK denote the set of all state vectors for the whole system for
which m k k
i=1 xi = lk , k = 1, 2, . . . , K . Assuming exchangeability of the
failure times of the mk components of type k
" K   #
Y mk −1 X
Φ(l1 , . . . , lK ) = × φ(x)
lk
k=1 x∈Sl1 ,...,lK

(UTOPIAE) Introduction to Reliability Theory 16 / 21

Probability system functions at time t

Let Ctk ∈ {0, 1, . . . , mk } denote the number of components of type k in

the system that function at time t > 0. If the probability distribution for
the failure time of components of type k is known and has CDF Fk (t),
and we assume failure times of components of different types to be
independent, then the probability that the system functions at time
t > 0 is
m1
X mK
X K
\
P(TS > t) = ··· Φ(l1 , . . . , lK )P( {Ctk = lk }) =
l1 =0 lK =0 k=1
m1 mK K 
"  #
X X Y mk
··· Φ(l1 , . . . , lK ) [Fk (t)]mk −lk [1 − Fk (t)]lk
lk
l1 =0 lK =0 k=1

(UTOPIAE) Introduction to Reliability Theory 17 / 21

Presented at 1st UTOPIAE Training School (Glasgow, November
2017).
Main challenges include upscaling to very large real-world systems
(with ‘zooming in’), applications to systems with multiple functions and
to networks with different routes through them.
For some questions of practical relevance other summaries of the full
structure function may be required.

(UTOPIAE) Introduction to Reliability Theory 18 / 21

Imprecision

Use of imprecise probability theory (also shown at 1st UTOPIAE

Training School) is attractive in reliability, particularly when considering
new or updated systems or components, or when there is doubt about
some model assumptions.
Inference and decision problems with imprecise probability tend to
require solution of constrained optimisation problems, with the
inference as target function and the set of probabilities as constraint.
There are more opportunities to use idea of imprecision to provide
robustness, enabling simple models to be used (e.g. in Accelerated
Life Testing).
Overall goal: Make the jobs of practitioners easier or better, preferably
both!
Many opportunities and challenges!

(UTOPIAE) Introduction to Reliability Theory 19 / 21

Resilience

In recent years there has been increasing attention to ‘Resilience’: if

things go wrong, how can negative effect be minimized?
Hugely important topic! For example for systems with multiple
functions or multiple phases, some critical functions or phases may still
be possible after a failure, possibly by re-configuring the system.
Of course, this also involves the usual aspects of maintenance, repair,
etc.
Very many challenges, perhaps generalizing structure function as
(imprecise) probability will prove useful.

(UTOPIAE) Introduction to Reliability Theory 20 / 21

The topic of ‘Reliability Theory’ is extremely wide, including aspects
from very many traditional areas of mathematics and engineering.

For substantial progress it is important to develop theory and methods

closely linked to real-world problems, as the gap between text-book
problems and reality is often very large.

UTOPIAE ESRs are well placed to make good contributions!

Feel free to contact me for more information or references.

(UTOPIAE) Introduction to Reliability Theory 21 / 21