Chapter 24 Planning Accelerated Destructive Degradation Tests William Q. Meeker and Luis A. Escobar Iowa State University and Louisiana State University Copyright 2002-2003 W. Q. Meeker and L. A. Escobar. Complements to the authors’ text Statistical Methods for Reliability Data, John Wiley & Sons Inc., 1998. Work done jointly with Danny L. Kugler and Laura L. Kramer (Imaging & Printing Group, Hewlett-Packard) January 13, 2014 3h 42min 24 - 1 Planning Accelerated Destructive Degradation Tests (ADDT) Chapter 24 Objectives • Outline practical issues in planning ADDTs. • Describe criteria for ADDT planning. • Illustrate how to evaluate the properties of a ADDTs. • Describe methods of constructing and choosing among ADDT plans. • Present guidelines for developing practical ADDT plans with good statistical properties. 24 - 2 Reasons for Conducting an Accelerated Test Accelerated destructive degradation tests (ADDTs) are used for different purposes. These include: • To identify failure modes and other weaknesses in product design. • Improve reliability. • Assess the durability of materials and components. • Monitor and audit a production process to identify changes in design or process that might have a seriously negative effect on product reliability. 24 - 3 Motivation/Example Reliability assessment of an adhesive bond (AdhesiveBondB) • Need: Estimate of the B05 and the proportion failing after 3 years (156 weeks) and 5 years (260 weeks) operating at room temperature of 25◦C. • Destructive measurements of strength in units of pounds. • Constraints ◮ 88 test units. ◮ 16 weeks for testing. • Test at 25◦C expected to yield little useful data. 24 - 4 A Class of Linear Degradation Models • These models are of the form yijk = µij + ǫijk = β0 + β1 exp(β ′2xj )τi + ǫijk where yijk , τi, and xj may be transformations of the measured degradation, ti, and the accelerating variable(s) AccVarj , respectively. • The model is linear in the sense that for specified AccVar vector xj , the degradation is linear in τi. ◮ For a scalar xj , β ′2xj = β2xj . ◮ For multiple AccVar, xj = (xj1, . . . , xjp)′, β ′2 = (β2, β3, . . . ), and β ′2xj is a linear combination of the AccVar, i.e., β ′2xj = p X i=2 βixji = β2xj2 + β3xj3 + · · · + βpxjp. 24 - 5 Degradation Model for the AdhesiveBondB ADDT • The degradation model for the planning and data analysis is yijk = β0 + β1 exp(β2xj )τi + ǫijk where yijk = log(Newtonsijk ) p √ τi = ti = Weeksi 11605 xj = − ◦ Cj + 273.15 (ǫijk /σ) ∼ Φnor (z). • Thus transformed time is square–root of Weeks and transformed AccVar is Arrhenius-transformed temperature. 24 - 6 Required Planning Information The ADDT planning requires some information which includes: • Purpose of the study. • Planning values for the parameters β that characterize degradation model for Dij . • A planning value for the parameter σ. • A plausible distribution to model the variability ǫ. • An specification of the critical degradation level Df . • The range of AccVar conditions available for experimentation. • The maximum length of the experiment. 24 - 7 Assumed Planning Information for the AdhesiveBondB Experiment The objective is to find a test plan to estimate with good precision the B05 and the proportion failing at use conditions (25◦C) after 3 and 5 years of operation. • A Normal failure-time distribution linear degradation model with constant regression parameters (β0, β1, β2) and same σ at each level of temperature. • Planning values for the parameters θ = (β0, β1, β2, σ)′ are β0✷ = 4.471, β1✷ = −864064160, β2✷ = 0.6364, σ ✷ = 0.1580 • Critical degradation level is specified as: Df = 40 pounds. Result: The planning information defines the degradation curves at all levels of temperature. It also defines the failure time distribution at a given temperature. 24 - 8 Alternative Method for Specifying the Planning Model Parameters • Specify planning values for β0, β2, σ, say β0✷ = 4.471 β2✷ = 0.6364 σ ✷ = 0.1580 • Specify the degradation rate (slope of the line), υ ✷ of µ(τ, x, β ) for a given temperature, say υ ✷ = −0.1025 at 50◦C. Because υ ✷ = β1✷ exp(β2✷x), with x = −11605/(50◦C + 273.15) = −35.912, then β1✷ = −0.1025 exp(0.6364 × 35.912) = −863499883 • The difference in the values for β1✷ obtained from the two procedures is due to rounding in the specification of υ ✷ and β2✷. 24 - 9 Degradation Paths from Planning Information D(τ, x, β ✷) = exp[β0✷ + β1✷ exp(β2✷x)τ ] √ τ = Weeks, x is Arrhenius-Transformed Temp Square Root–Log Axes 100 Newtons 50 20 70 DegreesC 60 DegreesC 50 DegreesC 10 0 10 20 30 40 50 60 Weeks 24 - 10 Original Test Plan for the ADDT AdhesiveBondB Data Test plan: 8 units with no ageing were measured at the start of the experiment. A total of 80 additional units were aged and measured according to the following temperatures and time schedule. Temp ◦C — 50 60 70 Totals 0 Weeks Aged 2 4 6 12 Totals 16 8 8 8 6 6 20 0 0 6 6 8 6 4 18 8 6 9 23 7 6 0 13 8 31 24 25 88 24 - 11 DegreesC Original AdhesiveBondB ADDT Plan 70 6 60 50 25 6 4 9 6 6 6 6 8 8 8 7 6 12 16 8 0 2 4 Weeks 24 - 12 Comments/Questions on the Original AdhesiveBondB ADDT Plan • The 8 observations at ti = 0 were not aged, i.e., there were never put into an oven. • Too much extrapolation in temperature? • Reasonable to extrapolate accurately to 25◦C? 24 - 13 Want a Plan That • Meets practical constraints and is intuitively appealing. • Is robust to deviations from assumed inputs. • Has reasonably good statistical properties. 24 - 14 Some Practical Guidelines for One AccVar ADDT Plans • Statistically optimum plans are a research topic. They are not practical plans but provide insight into the problem. • Use three or four levels of AccVar and spread them out as much as possible, subject to model adequacy. • Limit high level of the accelerating variable to a maximum reasonable condition. • Reduce lowest level of the accelerating variable (to minimize extrapolation)-subject to seeing some action. • Measure test units at three or four different times for each level of the AccVar. • Choose initial sample size using large-sample approximations. • Use simulation to compare alternative plans. 24 - 15 Evaluating Test Plan Properties Suppose inferences are needed on a function g(θ ) (one-toone and all the first derivatives with respect to the elements of θ exist, and are continuous). • Properties depend on test plan, model and (unknown) parameter values. Need planning values. b) • Large–sample approximate standard error of g(θ v" " # # u u ∂g(θ ) ′ ∂g(θ ) t b Ase[g(θ)] = Σb . θ ∂θ ∂θ where Σb is the inverse of the Fisher information matrix Iθ . θ For the θ = (β1, β 2, β3, σ)′ parameterization, the computation of Σb can be numerically unstable. Transform to stable θ parameters. • Monte Carlo simulation is a powerful tool. 24 - 16 Evaluating Test Plans • Large–sample approximations ◮ Fast. ◮ Provides some general insights. ◮ Provides simple sample size rules. • Simulation ◮ Reflects actual variability (no approximations). ◮ Provides visualization of sampling variability. ◮ Requires up-front computational effort. • Use large–sample approximations to get an initial indication of sample size needs and to suggest potential designs (e.g., use of optimization techniques). Use simulation to study the properties of particular designs. 24 - 17 Simulation of Original Test Plan Degradation 0.05 Quantile Versus Time Accelerated destructive degradation test simulation based on originalplan.ADDTplan AdhesiveBondB.Normal.ADDTpv 0.05 quantile of degradation versus Weeks at 25. DegreesC Resp:Log,Time:Square root,x:Arrhenius, Dist:Normal 72 68 Newtons 64 60 56 54 52 50 48 46 44 0 50 100 150 200 300 Weeks 24 - 18 Simulation of Original Test Plan FT (t) Estimate at 25◦C with Df = 40 Newtons Accelerated destructive degradation test simulation based on originalplan.ADDTplan AdhesiveBondB.Normal.ADDTpv Fraction failing versus Weeks for 40 Newtons at 25. DegreesC Resp:Log,Time:Square root,x:Arrhenius, Dist:Normal Lognormal Probability Plot .01 .005 .002 .001 Fraction Failing .0005 .0001 .00005 .00001 .000003 .000001 .0000003 .0000001 .00000003 .00000001 40 60 80 100 120 140 160 180 220 260 Weeks 24 - 19 Simulation of Original Test Plan Failure-Time 0.05 Quantile Estimates Versus Temperature Accelerated destructive degradation test simulation based on originalplan.ADDTplan AdhesiveBondB.Normal.ADDTpv Failure time 0.05 quantile vs DegreesC for failure definition 40 Newtons Resp:Log,Time:Square root,x:Arrhenius , Dist:Normal 10000 Weeks 1000 100 10 1 20 30 40 50 60 70 DegreesC 24 - 20 Simulation of Original Test Plan Joint Distribution of tb.05 and βb2 Estimates 0 5 15 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0 1000 2000 3000 4000 5000 8 12 t_0.05 at 25 DegreesC for a failure def of 40 Newtons 0.50 0.60 0.70 0.80 beta2 Mean of the simulation estimates 4 Focus quantity 1: 1588 beta2: 0.6509 0 beta2 Accelerated destructive degradation test simulation based on originalplan.ADDTplan AdhesiveBondB.Normal.ADDTpv Failure time 0.05 quantile vs DegreesC for failure definition 40 Newtons Resp:Log,Time:Square root,x:Arrhenius , Dist:Normal 0 1000 3000 t_0.05 at 25 DegreesC for a failure def of 40 Newtons 24 - 21 ADDT Sample Size Determination for a Quantile tp • Approximate 100(1 − α)% confidence interval for log(tp): " ! # r b ˜ p) = log(tbp)± √1 z = log(tbp)±log(R log(tp), log(t (1−α/2) Vlog(b t ) p n e Exponentiation yields a confidence interval for tp [tbp/R, tbpR] where √ r b R = exp (1/ n)z(1−α/2) V = log(b tp ) s t˜p/tp. e b d /tb2 with V✷ • Replace V = nVar , R with RT , and b log(b tp ) tp p log(b tp ) solve for n to get n= 2 z(1−α/2) V✷ log(b tp ) . 2 [log(RT )] 24 - 22 Statistical Properties of a Test Plan With a Given Sample Size n Test plans with a sample size of n= 2 z(1−α/2) V✷ log(b tp ) [log(RT )]2 provides confidence intervals for tp having the following characteristics: • In repeated samples approximately 100(1 − α)% of the intervals will contain tp. b b • In repeated samples V is random and if V > b log(tr log(b tp ) p) V✷ b then the ratio R = t˜p/tp will be greater than RT . log(tp ) • The ratio R = r e t˜p/tp will be greater than RT with a probe ability approximately equal to 0.5. 24 - 23 Sample Size to Estimate the AdhesiveBondB t0.05 Quantile at 25◦C when Df = 40 Newtons Needed sample size giving approximately a 50% chance of having a confidence interval factor for the 0.05 quantile that is less than R use condition= 25. DegreesC and a failure definition= 40 originalplan.ADDTplan AdhesiveBondB.Normal.ADDTpv 10000 5000 Sample Size 2000 1000 500 200 100 50 99% 20 95% 90% 80% 10 1.0 1.5 2.0 2.5 3.0 Confidence Interval Precision Factor R 24 - 24 A Two-Variables Linear Degradation Model With Temperature and Relative Humidity as AccVar • AdhesiveBondD model yijk = β0 + β1 exp(β ′2xj )τi + ǫijk = β0 + β1 exp(β2xj2 + β3xj3)τi + ǫijk where yijk = log(Newtonsijk ) p √ ti = Weeksi τi = (ǫijk /σ) ∼ Φnor (z). • The AccVar transformations are Arrhenius for Temp (Temperature) and logit for RH (Relative Humidity), i.e., 11605 xj2 = − ◦ Cj + 273.15 ! RHj xj3 = log . 1 − RHj 24 - 25 Original Test Plan for the AdhesiveBondD ADDT Data • A 8 × 3 × 2 full factorial arrangement in Weeks, Temp, and RH, with 6 units at each factor combination. RH % 20 Temp ◦C 50 60 70 80 50 60 70 Totals 0 1 6 6 6 6 6 6 36 6 6 6 6 6 6 36 Weeks Aged 2 4 6 8 6 6 6 6 6 6 36 6 6 6 6 6 6 36 6 6 6 6 6 6 36 6 6 6 6 6 6 36 Totals 12 16 6 6 6 6 6 6 36 6 6 6 6 6 6 36 48 48 48 48 48 48 288 • Might want to re-allocate units assigned to the corners of the experimental region, i.e., highest (lowest) levels of (Weeks, RH, Temp). Also, might want to decrease the number of pens allocated to each factor level combination to reduce size of the experiment. 24 - 26 Some Practical Guidelines for Two or More AccVar ADDT Plans • Moderate increases in two accelerating variables may be safer than using a large amount of a single accelerating variable. • There may be interest in assessing the effect of nonaccelerating variables, i.e., factor Weeks in the AdhesiveBondD example. • There may be interest in assessing joint effects of two or more accelerating variables, i.e., (RH, Temp) in the AdhesiveBondD example. • Traditional factorial experiments are good as starting plans. • When possible, a fractional factorial could provide economy in testing. • Choose initial sample size using large-sample approximations. • Use simulation to compare alternative plans. 24 - 27 Choosing Experimental Variable Definition to Minimize Interaction Effects • Care should be used in defining experimental variables. • Guidance on variable definition and possible transformation of the response and the experimental models should, as much as possible, be taken from mechanistic models. • Proper choice can reduce the occurrence or importance of statistical interactions. • Models without statistical interactions simplify modeling, interpretation, explanation, and experimental design. • Knowledge from mechanistic models is also useful for planning experiments. 24 - 28 Examples of Choosing Experimental Variable Definition to Minimize Interaction Effects • For accelerated testing of dielectrics, use size and volts stress (e.g., mm and volts/mm instead of mm and volts). • For light exposure, use aperture and total light energy (not aperture and exposure time). • For accelerated humidity testing with a corrosion mechanism, use RH and temperature (not vapor pressure and temperature). 24 - 29 • • • • • • • 100 150 200 3.0 • • • 2.0 • • • 50 Voltage Stress (Volts/mm) 1.0 1.5 Size (mm) 2.0 1.5 1.0 Size (mm) • 2.5 • 2.5 3.0 Comparison of Experimental Layout with Volts/mm Versus Size and Volts Versus Size • 100 • • 300 500 Volts 24 - 30 • • • • • • • 200 300 3.0 • • • 2.0 • • • 50 100 Volts 1.0 1.5 Size (mm) 2.0 1.5 1.0 Size (mm) • 2.5 • 2.5 3.0 Comparison of Experimental Layout with Volts versus Size and Volts/mm versus Size • • • 100 200 300 400 Voltage Stress (Volts/mm) 24 - 31 Areas for Future Research in ADDT Planning • Optimum plans (to provide insight and general rules). • Optimized compromise plans (using practical constraints). • Plans that allow the use of prior information on activation energy (i.e., plans for Bayesian analysis). • Plans allowing for censoring and coarse data (accounting for measurement limitations). 24 - 32 Technical Details The following slides give technical details used in SPLIDA to implement the methodology. 24 - 33 ADDT Model Asymptotic Variances Under certain regularity conditions the following results hold asymptotically (large sample) b ∼ • θ ˙ MVN(θ , Σb ), where Σb = Iθ−1, and θ θ " Iθ = E − ∂ 2L(θ ) ∂ θ∂ θ′ # " X ∂ 2Lij (θ ) =n πij E − ′ ∂ θ ∂ θ ij # where πij is the proportion of observations allocated to (ti, xj ). b) ∼ • For a scalar gb = g(θ ˙ NOR[g(θ ), Avar(gb)], where Avar(gb) = " ∂g(θ ) ∂θ #′ " # ∂g(θ ) . θ ∂θ Σb • When g(θ ) is positive for all θ , then b )] ∼ log[g(θ ˙ NOR{log[g(θ )], Avar[log(gb)]}, where Avar[log(gb)] = 1 g !2 Avar(gb). 24 - 34 Asymptotic Approximate Standard Errors for a Function of the Parameters g(θ ) Given a specified model and parameter values (but without need to specify sample size), one can compute scaled asymptotic variances. • The variance factors Vgb = nAvar(gb) and Vlog(gb) = nAvar[log(gb)] may depend on the actual value of θ but they do not depend on n. To compute these variance factors one uses planning values for θ (denoted by θ ✷) as discussed later. • The asymptotic standard error for gb and log(gb) are 1 q Ase(gb) = √ Vgb n q 1 Vgb 1 q Vlog(gb) = √ . Ase[log(gb)] = √ n n g • Easy to choose n to control Ase. 24 - 35 ADDT Model Fisher Information Matrix • The Fisher information matrix, Iθ , is " ∂ 2L # X " ∂ 2L ij =n πij E − ∂ θ∂ θ′ ∂ θ∂ θ′ ij X = n πij Iij Iθ = E − # ij where L = log[L(θ )], Lij is the contribution of a single observation at ξ ij = (ti, xj ) to the log-likelihood, and πij is the proportion of observations made at ξ ij . E is the expectation operator, the index on the summation operation runs over the distinct time, and stress level combinations ξ ij . Iij is the contribution of one observation at ξ ij to Iθ . 24 - 36 Contributions to the ADDT Model Fisher Information Matrix • The contribution Iij to the Fisher information matrix is ′ 1 f11(ξ ij )uij uij Iij = 2 σ symmetric f12(ξ ij )uij f22(ξ ij ) where and f11(ξ ij ), f12(ξ ij ), f22(ξ ij ) are the LSINF elements for the distribution Φ at the experimental conditions ξ ij and uij is the vector of partial derivatives of the degradation path with respect to the β parameters uij = ∂µij ∂β0 ∂µij ∂β1 ∂µij ∂ β2 1 ′ x )τ = exp( β 2 j i β1 xj exp(β ′2xj )τi . • For test planning the Fisher matrix is evaluated at planning values θ ✷. 24 - 37 A Reparameterization of the Linear ADDT Model for Numerical Stability • The model is as before yij = β0 + β1 exp(β ′2xj )τi + σǫij • Suppose that x̄ is the centroid of the stress variables [i.e., x̄ = (x̄1, . . . , x̄k )′] and τ̄ is an average transformed time. Then the model can be reparameterized as yij = γ0 + γ1 exp ′ xj − x̄ γ 2 τi − τ̄ + σǫij where γ0 is the intercept for the average stress line (i.e., degradation line for x̄) at τ̄ ; γ1 is the slope of the average stress line; and γ 2 = β 2 are the regression coefficients corresponding to the x variables. • It can be shown that γ1 is the geometric mean of the slopes β1 exp(β ′2xj ), for the values of xj stress variables in the data set. • The vector ϕ = (γ0, γ1, γ 2, σ)′ denotes the stable parameters. 24 - 38 Relationship Between the Stable Parameters ϕ and the Original Parameters θ • It can be shown that γ0 = β0 + β1 exp(β ′2x̄)τ̄ γ1 = β1 exp(β ′2x̄) γ 2 = β 2. • Solving for the βs β2 = γ 2 β1 = γ1 exp(−x̄′γ 2) β0 = γ0 − γ1τ̄ . 24 - 39 ADDT Fisher Matrix (Transformed Time Scale) • The contribution Iij to the Fisher information matrix is ′ 1 f11(ξ ij )v ij v ij Iij = 2 σ symmetric f12(ξ ij )v ij f22(ξ ij ) where and f11(ξ ij ), f12(ξ ij ), f22(ξ ij ) are the LSINF elements for the distribution Φ at the experimental conditions ξ ij and v ij is the vector of partial derivatives of the degradation path with respect to the γ = (γ0, γ1, γ 2) parameters v ij = ∂µij ∂γ0 ∂µij ∂γ1 ∂µij ∂γ2 1 i h ′ = exp γ 2(xj − x̄) τi − τ̄ i h γ1 (xj − x̄) exp γ ′2(xj − x̄) τi . • For test planning, the Fisher matrix is evaluated at the planning values θ ✷. 24 - 40 Alternative Expression for the Variance-Covariance Matrix • Let Σγb be the variance-covariance matrix in the stable parameterization and Σb the corresponding matrix in the origθ inal parameterization. • Then Σb = AΣγb A′, where θ A = = ∂β0 ∂β0 ∂β0 0 ∂γ0 ∂γ1 ∂ γ 2 ∂β1 ∂β1 ∂β1 0 ∂γ0 ∂γ1 ∂ γ 2 ∂ β2 ∂ β2 ∂ β2 0 ∂γ0 ∂γ1 ∂ γ 2 0 0 0′ 1 1 −τ̄ 0′ 0 exp[−γ ′2x̄] −γ1x̄′ exp[−γ ′2x̄] 0 0 I 0 0 0′ 0 0 0 1 0 is a vector of zeros and I is an identity matrix with the same row dimension as x. 24 - 41 Asymptotic Variance of ht(tbp) ν • For decreasing degradation ht(tp) = ν+ςΦ−1(p), for p > Φ − . ς Then ignoring the probability spike, we get 2 Thus ∂ht (tbp) Avar[ht(tbp)] = ∂ tbp t✷ p ∂h−1 (z) t Avar(tbp) = ∂z Avar(tbp). 2 ht (t✷ p) Avar[ht(tbp)]. • This allows to write Avar[ht(tbp)] as a function of Σb or as θ a function of Σϕ b. 24 - 42 Asymptotic Variance of ht(tbp) in Funtion of the θ and ϕ Parameters. • Define ∂ht(tp) cϕ = ∂ϕ It can be shown that cϕ = ∂θ ∂ϕ !′ and ∂ht(tp) cθ = . ∂θ " # ∂ht(tp) ∂ht(tp) = A′ = A ′ cθ . ∂θ ∂θ • Then direct computations yield Avar(tbp) = = = !2 −1 ∂ht (z) c′θ Σθb cθ ∂z !2 −1 ∂ht (z) ′ c′θ AΣϕ b A cθ ∂z !2 −1 ∂ht (z) c′ϕ Σϕ b cϕ . ∂z 24 - 43 Standardized Asymptotic Variance Formula • It can be shown that n σ2 ∂ht(z) ∂z !2 Avar(tbp) = n σ2 c′ϕ Σϕ b cϕ −1 ′ = c ϕ F ϕ (ξ ) cϕ where f11(ξ ij )v ij v ′ij X F ϕ (ξ ) = πij symmetric ij f12(ξ ij )v ij f22(ξ ij ) . • The standardized information matrix Fϕ(ξ ) depends only on the test plan (ξ ij , πij ), ϕ, and the derivatives v ij . 24 - 44 Derivatives • For the square root of time, ht(t) = −1 ∂ht (z) ∂z 2 = = 2z ∂z ∂z √ t, transformation, = 2ht(t✷ p ). ht (t✷ p) • The elements of cθ are: ∂ht(tp) 1 = − ∂β0 β1 exp(β ′2x) ht(tp) ∂ht(tp) = − ∂β1 β1 ∂ht(tp) = − x ht(tp) ∂β2 ∂ht(tp) Φ−1(p) = − ∂σ β1 exp(β ′2x) 24 - 45 Derivatives-Continued • The elements of cϕ are: ∂ht(tp) ∂γ0 ∂ht(tp) ∂γ1 ∂ht(tp) ∂γ2 ∂ht(tp) ∂σ 1 = − γ1 exp[γ ′2(x − x̄c)] # " 1 τ̄c − ht(tp) = ′ γ1 exp[γ 2(x − x̄c)] = − (x − x̄c)ht(tp) = − 1 γ1 exp[γ ′2(x − x̄c)] where (x̄c, τ̄c) is a centroid of the data. • The elements of cθ and cϕ were obtained directly. A check for the expressions is cϕ = A ′ cθ 24 - 46 Linearization of the Degradation Model • A first order approximation about ϕ✷ gives zijk = γ0 + γ1 = γ0 + γ1 where zijk ✷ v1ij ✷ v2ij ∂µij ∂µij ✷ ✷ − γ1 + γ2 − γ2 + σǫijk ∂γ1 ∂γ2 ✷ + γ − γ ✷ v ✷ + σǫ − γ1✷ v1ij 2 ijk 2 2ij n h i o ✷ = yijk − exp γ2 xj − x̄c τi − τ̄c i h ✷ = exp γ2 (xj − x̄c) τi − τ̄c i h ✷ ✷ = γ1 (xj − x̄c) exp γ2 (xj − x̄c) τi • The asymptotic variance for this model is f11(ξ ij )v ij v ′ij (σ ✷)2 X πij Σϕ b = n ij symmetric i−1 (σ ✷)2 h , F ϕ ✷ (ξ ) = −1 f12(ξ ij )v ij f22(ξ ij ) n ✷ , v ✷ )′ and ϕ✷ = (1, γ ✷ , γ ✷ , γ ✷ , σ ✷ )′ . where v ij = (1, v1ij 0 1 2 2ij 24 - 47 Optimal Test Plans • Statistically optimum plans are a current research topic. They are not practical plans but provide insight into the problem. • Statistical optimum plans can be obtained numerically under the following criteria: i−1 h ′ argξ,π min Avar(tbp; ξ , π ) = argξ,π min cϕ✷ Fϕ✷ (ξ ) c ϕ✷ . Notice that the derivatives, c and the information matrix, F , are evaluated at ϕ✷. • The General Equivalence Theorem can be used to check the optimality of the test plan obtained from the numerical algorithm. 24 - 48 The General Equivalence Theorem to Check Optimal Test Plans • Defining ξ v as a plan with all units at v and dc(ξ , v ) = cϕ ′[Fϕ(ξ )]−1Fϕ(ξ v )[Fϕ(ξ )]−1 cϕ−cϕ ′ [Fϕ(ξ )]−1 cϕ where c and Fϕ are evaluated at ϕ✷. • The test plan (ξ copt, πcopt) minimizes Avar(tbp; ξ , π ) iff sup dc(ξ copt, x) = 0. v 24 - 49 General Guidelines • Reduce lowest level of the accelerating variable (to minimize extrapolation)-subject to seeing some action. • Measure test units at three or four different times for each level of the AccVar. • Choose initial sample size using large-sample approximations. • Use simulation to compare alternative plans. 24 - 50 Standardization of Explanatory Variables • Simple scaling of the explanatory variables and time gives yijk = β0s + β1s exp[(β s2)′wj ]vi + σǫijk where wj are the scaled explanatory variables and vi is the scaled time variable (all explanatory variables scaled in [0, 1]). • When xj = xj is a scalar, then xj − xL , 0 ≤ wj ≤ 1. wj = xU − xL τi vi = , 0 ≤ vi ≤ 1. τH β0s = β0 β1s = τL exp[β2xL] β1 β2s = (xH − xL)β2 xL, xH are the smallest and largest possible value of the explanatory variable x and τH is the largest possible time in the experiment. 24 - 51