Skip to content

Reliability and Operations Research Formula Sheet and Short Notes

One-page CSIR NET revision: hazard and reliability, censored life tests, system reliability, simplex and duality, M/M/1 to M/G/1 queues and EOQ models.

  • 4 min read
  • Updated
On this page · 8 sections
  1. The paper in one table
  2. Lifetimes and hazard
  3. Censoring and life testing
  4. System reliability
  5. Linear programming
  6. Queues
  7. Inventory
  8. Part C checklist

Every formula and theorem from Hazard Function, Failure Rates, Censoring and Life Testing, Series and Parallel Systems, Linear Programming: Simplex and Duality, Queuing Models and Inventory Models, on one page.

The paper in one table

Per the NTA information bulletin for December 2026 (Mathematical Sciences, code 704):

PartGivenAttemptCorrectWrongMaximum
B4025+3−0.7575
C6020+4.75095

Part C: one or more options may be correct; marks only if you mark all correct options and no wrong one. No partial credit, no negative marking. Details on the Mathematical Sciences overview.

Lifetimes and hazard

$$h(t) = \frac{f(t)}{R(t)} = -\frac{d}{dt}\ln R(t), \qquad \boxed{R(t) = e^{-\int_0^t h(u)\,du}}, \qquad \text{MTTF} = \int_0^\infty R(t)\,dt, \qquad m(t) = \frac{\int_t^\infty R}{R(t)}.$$
Distribution\( R(t) \)\( h(t) \)Class
Exponential (mean \( \theta \))\( e^{-t/\theta} \)\( \frac{1}{\theta} \)constant; memoryless; \( m(t) = \theta \)
Weibull\( e^{-(t/\eta)^\beta} \)\( \frac{\beta}{\eta}(\frac{t}{\eta})^{\beta - 1} \)\( \beta < 1 \) DFR, \( = 1 \) exp, \( > 1 \) IFR; mean \( \eta\Gamma(1 + \frac{1}{\beta}) \)
Rayleigh\( e^{-ct^2/2} \)\( ct \)IFR
Gamma, shape \( k \)\( \to \lambda \)\( k > 1 \) IFR, \( k < 1 \) DFR
Uniform \( (0, b) \)\( 1 - \frac{t}{b} \)\( \frac{1}{b - t} \)IFR, unbounded hazard
Lomax\( (1 + \frac{t}{\sigma})^{-\alpha} \)\( \frac{\alpha}{\sigma + t} \)DFR
Mixture of exponentialsdecreasingDFR
  • Classes: IFR \( \Rightarrow \) IFRA \( \Rightarrow \) NBU \( \Rightarrow \) NBUE; IFR \( \Rightarrow \) DMRL \( \Rightarrow \) NBUE.
  • Bathtub: infant mortality (DFR), useful life (constant), wear-out (IFR). Burn-in helps only under DFR.
  • Average failure rate on \( [t_1, t_2] \): \( \frac{\ln R(t_1) - \ln R(t_2)}{t_2 - t_1} \).

Censoring and life testing

SchemeFixedRandom
Type Itest time \( t_0 \)number of failures
Type IInumber of failures \( r \)test time \( t_{(r)} \)
Randomcensoring times \( C_i \)
  • Likelihood: \( \prod f(t_i)^{\delta_i} R(t_i)^{1 - \delta_i} \).
  • Exponential: \( \hat\theta = \frac{T}{r} \), \( T \) = total time on test.
  • TTT: without replacement \( \sum_{i \le r} t_{(i)} + (n - r)t_{(r)} \) (Type I: \( (n - r)t_0 \)); with replacement \( n t_{(r)} \) (Type I: \( n t_0 \)).
  • Type II: \( \frac{2T}{\theta} \sim \chi^2_{2r} \); \( \hat\theta \) unbiased, \( \text{Var} = \frac{\theta^2}{r} \); CI \( \left(\frac{2T}{\chi^2_{2r, 1 - \alpha/2}}, \frac{2T}{\chi^2_{2r, \alpha/2}}\right) \).
  • Expected duration (without replacement): \( \theta\sum_{i=1}^r \frac{1}{n - i + 1} \); with replacement \( \frac{r\theta}{n} \).
  • Kaplan-Meier: \( \hat R(t) = \prod_{t_j \le t}(1 - \frac{d_j}{n_j}) \); Nelson-Aalen \( \hat H = \sum \frac{d_j}{n_j} \). Censored units leave the risk set after their time; \( \hat R \) jumps only at failures.

System reliability

System\( R_S \)MTTF (identical exponential, rate \( \lambda \))
Series\( \prod p_i \); hazards add\( \frac{1}{\sum \lambda_i} \)
Parallel\( 1 - \prod q_i \)\( \frac{1}{\lambda}(1 + \frac{1}{2} + \dots + \frac{1}{n}) \)
Two parallel, rates \( \lambda_1, \lambda_2 \)\( e^{-\lambda_1 t} + e^{-\lambda_2 t} - e^{-(\lambda_1 + \lambda_2)t} \)\( \frac{1}{\lambda_1} + \frac{1}{\lambda_2} - \frac{1}{\lambda_1 + \lambda_2} \)
k-out-of-n\( \sum_{j \ge k}\binom{n}{j}p^jq^{n-j} \)\( \frac{1}{\lambda}\sum_{i=k}^{n}\frac{1}{i} \)
2-out-of-3\( 3p^2 - 2p^3 \)\( \frac{5}{6\lambda} \) (less than one unit)
Cold standby (perfect switch)\( e^{-\lambda t}\sum_{j\( \frac{n}{\lambda} \)
Bridge (all \( p \))\( 2p^2 + 2p^3 - 5p^4 + 2p^5 \)
  • Pivotal decomposition: \( R_S = p_i R_S(1_i) + q_i R_S(0_i) \).
  • Bounds: \( \prod_{\text{cuts}}(1 - \prod q_i) \le R_S \le 1 - \prod_{\text{paths}}(1 - \prod p_i) \).
  • Birnbaum importance \( \frac{\partial R_S}{\partial p_i} \): series, the weakest component; parallel, the strongest.
  • Series of IFR is IFR; parallel of IFR need not be (two unequal exponentials); coherent systems preserve IFRA.
  • Component-level redundancy \( \ge \) system-level redundancy.

Linear programming

ItemFact
Geometryfeasible set convex; vertices = BFS; at most \( \binom{n}{m} \); optimum (if any) at a BFS; optimal set convex
Simplex (max)enter: most negative \( z_j - c_j \); leave: min \( \frac{b_i}{a_{ik}} \), \( a_{ik} > 0 \); optimal when all \( z_j - c_j \ge 0 \)
Alternative optimanon-basic \( z_j - c_j = 0 \)
Unboundedentering column has no positive entry
Infeasibleartificial positive at the end (Big-M or Phase I \( w^* > 0 \))
Degeneratebasic variable 0; possible cycling; Bland’s rule prevents it
Dual pair\( \max c^Tx, Ax \le b, x \ge 0 \ \leftrightarrow\ \min b^Ty, A^Ty \ge c, y \ge 0 \)
Sign rules (max primal)\( \le \to y \ge 0 \); \( \ge \to y \le 0 \); \( = \to y \) free; \( x \ge 0 \to \ge \); \( x \le 0 \to \le \); \( x \) free \( \to = \)
Weak / strong\( c^Tx \le b^Ty \); equal optimal values
Outcomesboth optimal; unbounded / infeasible; both infeasible (never both unbounded)
Complementary slackness\( y_i \times \text{slack}_i = 0 \), \( x_j \times \text{surplus}_j = 0 \)
Shadow prices\( z \)-row under slack columns; valid while \( B^{-1}b \ge 0 \)

Benchmark: \( \max 3x_1 + 2x_2 \), \( 2x_1 + x_2 \le 8 \), \( x_1 + 2x_2 \le 7 \), \( x_2 \le 3 \): optimum \( (3, 2) \), \( z^* = 13 \), duals \( (\frac{4}{3}, \frac{1}{3}, 0) \).

Queues

$$L = \lambda W, \quad L_q = \lambda W_q, \quad W = W_q + \frac{1}{\mu}; \qquad p_n = p_0\prod_{i=1}^{n}\frac{\lambda_{i-1}}{\mu_i}.$$
ModelResults
M/M/1 (\( \rho < 1 \))\( p_n = (1 - \rho)\rho^n \); \( L = \frac{\rho}{1 - \rho} \); \( L_q = \frac{\rho^2}{1 - \rho} \); \( W = \frac{1}{\mu - \lambda} \); \( W_q = \frac{\rho}{\mu - \lambda} \); \( P(N \ge n) = \rho^n \)
M/M/1/K (any \( \rho \))\( p_n = \frac{(1 - \rho)\rho^n}{1 - \rho^{K+1}} \) (\( \frac{1}{K + 1} \) if \( \rho = 1 \)); \( \lambda_{\text{eff}} = \lambda(1 - p_K) = \mu(1 - p_0) \)
M/M/c (\( \lambda < c\mu \))\( p_0^{-1} = \sum_{n
Erlang C\( P(\text{wait}) = \frac{a^c p_0}{c!(1 - \rho)} \)
M/M/c/K (any \( \rho \))\( L_q = \frac{p_0 a^c\rho}{c!(1 - \rho)^2}[1 - \rho^{K-c+1} - (K - c + 1)(1 - \rho)\rho^{K-c}] \)
Erlang B (M/M/c/c)\( \frac{a^c/c!}{\sum_{j \le c}a^j/j!} \)
M/G/1 (P-K)\( L_q = \frac{\lambda^2\sigma^2 + \rho^2}{2(1 - \rho)} \); \( W_q = \frac{\lambda E[S^2]}{2(1 - \rho)} \); M/D/1 = half M/M/1

Benchmarks (\( \lambda = 6 \)): M/M/1 with \( \mu = 8 \): \( L = 3 \), \( L_q = 2.25 \), \( W = 30 \) min, \( W_q = 22.5 \) min. M/M/2 with \( \mu = 4 \): \( p_0 = \frac{1}{7} \), \( P(\text{wait}) = \frac{9}{14} \), \( L_q = \frac{27}{14} \), \( W = \frac{4}{7} \) h.

Inventory

Model\( Q^* \)\( TC^* \)
EOQ\( \sqrt{\frac{2AD}{h}} \)\( \sqrt{2ADh} \)
Backorders (cost \( s \))\( \sqrt{\frac{2AD}{h}\cdot\frac{h + s}{s}} \); \( S^* = \frac{s}{h + s}Q^* \)\( \sqrt{2ADh\cdot\frac{s}{h + s}} \)
Production rate \( P \)\( \sqrt{\frac{2AD}{h(1 - D/P)}} \); \( I_{\max} = Q(1 - \frac{D}{P}) \)\( \sqrt{2ADh(1 - \frac{D}{P})} \)
Newsvendor\( F(Q^*) = \frac{c_u}{c_u + c_o} \)
  • At the EOQ, ordering cost = holding cost. \( \frac{TC}{TC^*} = \frac{1}{2}(r + \frac{1}{r}) \), \( r = \frac{Q}{Q^*} \): a factor 2 costs 25%.
  • Reorder point \( DL \) (inventory position). Discounts: candidates are in-range EOQs and break points; compare total cost including \( Dc \).

Part C checklist

Back to the Reliability and Operations Research chapter overview. For the Markov chain theory behind the queues, see Markov Chains and Classification of States.

Advertisement
Advertisement