Modern Spares Analysis - PowerPoint Presentation

Modern Spares AnalysisFor Non-Specialists Robert Butler Presented at LOA University, 8 October 2018, Oklahoma City

Analysis of Spare Stocks Fundamental ideasAn example problemStock optimizationAccounting for Time

The Purpose of Spare Stocks Only a working equipment or system can provide utility, not the resources required to fix itThe quantity of a system’s utility actually delivered is a joint function ofThe intrinsic benefits derived from having the use of the system – performanceThe amount of performance delivered or deliverable – operational availability or Ao Stocks of spares, along with several other support resources, increase Ao and hence, the rate at which utility is supplied Although any increase in stock will increase A o , maximum increase requires economic optimization So we care about Enough stock to achieve a performance goal Since budgets are fixed, we want to do this at the lowest possible cost

Stocks Versus Flows The stock is the amount of water in the bathtub, the flow is the rate at which it leaves Notice that the stock remains the same only if Flow 1 = Flow 2 Flow 1 Flow 2 A o

Alternative Stock CalculationMethods Single item methodsRules of thumb Arbitrary percentage that seemed about right in the pastExpert opinion: someone who knows a lot about the system guesses how many sparesSaw tooth deterministic inventory models Simple relationships with uncertainty – Poisson assumptionConstant k models ( )USN’s FILSIP Item fill rate (probability of no stock out at the base) System methods More satisfactory because they focus on delivery of utility Require knowledge of the fleet being spared Not necessarily optimized Steady state system optimization USAF/USN/US Army models all use VARI-METRIC approach based on work originally done by Craig Sherbrooke at RAND and later, LMI New issues: dealing with timeOvercoming the limitations of steady state assumptionsApplication of optimizing techniques to day-to-day spares management

The Theory versus What the Item Manager Sees Q R t s Data for each item include Q (order quantity) R (re-order point) demand rate foreseen events Dues-in from repair or replenishment Tests Shelf life expirations The simulation starts from the current position and projects it forward Each subsequent simulation repeats this with increased R

Stock Optimization Mix of partsBuy more of the less expensive parts in substitution for the more expensive parts, other things equalBuying lower-indenture items will reduce the number of higher (more expensive) items required Geographic location of stocksThe closer a part is located to the operating equipmentThe greater its impact on equipment at that location The less its impact on all other equipmentThe higher the echelon at which a stock is keptThe smaller its impact on any specific equipment The greater its contribution to all equipments

General Optimization Process System LRU 1 LRU 2 LRU 3 LRU 4 LRU 5 LRU 6 Step 1: Choose item with highest ratio Step 2: Re-compute ratio for that item Step 3: Repeat steps 1 and 2 until target reached

The Effect of Multiple Indentures and Locations/Echelons Complex hardware breakdown structures require testing the effect of buying lesser parts to replace their parent assembliesProblems arise such asA part is an LRU in one application and not in another An LRU is physically part of an SRU assembly (spark plug)The part search must be done for every locationOther complicationsEach location may have different operating programs, delay times, stock constraints and MOE requirement

Other Complicating FactorsWeight and volume constraints and costsNASA requirement to account for up-mass Sparing helicopters on board small shipsSolution is the use of shadow pricing – setting a price per unit of weight or volumeRedundancy calculationsK of n redundancy is common in some technologies Multiple pricesBuying versus sellingRepair from unserviceable inventoryNon-steady state considerationsPredictable changes (reliability, lead time …) Foreseeable or planned changes (basing, fleet size…)

Operation of a Spares ModelMarginal Optimization Each point is another calculation of max “bang/buck” at the margin The yellow line is the locus of optimal solutions – nothing above is feasible, all below is inferior

Three Case StudiesThe (very) first case: George AFB, 1965-1966 Canadian airlinesSeveral commercial airline fleets

George AFB Study, 1965-6The Original METRIC Model Operations monitored for 6 monthsFlying hour programNot operationally ready rates (NORS rates) compiledRe-stocked in middle of periodRemoved all existing base stock Re-stocked with METRIC recommendations based on equal fill rateResultsSlight drop in NORS ratesHalf the cost of stock

Canadian Airlines Pilot project on A320 fleet using VMetricStarting inventory of $31 millionStarting service level (fill rate) of 70% Final inventory of $19.4 million $18.6 million excess sold$7 million shortages acquiredFinal service level of 85%Study required 60 days Used simulation model to prove solution correctnessAdopted technique for all fleets – savings of $80 million out of $200 million starting inventory

Orange: increase in service level Blue: inventory reduction value Optimization WorksThe Case of Commercial Airline Fleets In all cases, maximum inventory reduction at equal service level would have been 50% or greater.

What About Time? Multi-period optimization and Life Cycle Inventory Cost Tactical Inventory Optimization

Accounting for Time in Spares OptimizationSteady state means: All inputs assumed to remain true indefinitelyMOEs are true for average of all time, not each periodSeveral problems are defined by timeLong lead time versus short lead time parts Phase-in and phase-out of fleetsHandling obsolescence and DMSMSEnd of system life and life extensionChanging system configurationMRO and parts supply delay changesPart attribute changes from expected values To name just a few…

Two Ways to Deal With TimeSpare parts planningParts planners need optimizing models that recognize and deal with time Spare parts managementParts managers need optimized advice for day-to-day actionsThis depends on what is where compared to what should be whereWhich, in turn, depends on a complete model of the system’s operation and maintenance over time

Spares Optimization Over Multiple Periods Multi-period spares analysis can overcome some of the problems of steady state models such as:Fleet build up and run down Long lead time vs. short lead time optimal choiceDMSMS, obsolescence and technology insertion issues Mid-life upgrade calculationsAging and end-of-life fleet sparing NOTE: An optimization algorithm that translates EBO into a monetary metric can also optimize spares for complex mixtures of metrics like A o and fill rate – often found in contractual incentive clauses.

Predictable vs. Foreseeable ChangesPredictable changes can be ascribed to parts or part groups Lead time to procure or repair partsProportion of parts subject to obsolescenceAverage useful life (mean technological life, MTL) and expired life Reliability improvementConfiguration changesForeseeable changes are in the hands of planners Fleet build-up, run-downBasing changesTemporary deploymentsOperating pace changes Lead time improvements

The Difference Between Steady State and Time-Sensitive Models The bang-for-buck ratio at the heart of the optimization process must now be changed to recognize time The “timeless” version of EBO is replaced by the discounted present monetary value of the stream of EBO reductions available from this part Steady State Time-Sensitive

Life Cycle Inventory Cost (LCIC) The first period solution in a multi-period spares optimization will be more costly than a steady-state solutionBecause, among the cheaper (myopic) steady-state choices, some will lose their utility in “less than forever” End of system life Technological or market-driven obsolescenceMajor configuration changes (intended obsolescence) To understand full benefit, it is necessary to view inventory from a life cycle perspective

The Essence of LCIC Time-Sensitive Solution Steady State Solution Right shift of curve is cost of new part number stock

Tactical Spares OptimizationDay-to-Day Guidance for Item Managers Monitoring all inventory and maintenance transactions, determine What is Where (WW)Using system simulation, determine What Will Be Where (WWW) over the near future and what backorder risks are associated with this Using a deterministic optimization model, calculate What Should Be Where, WSWUsing a simple cost trade-off model determine the most cost effect tactics required to convert WWW to WSW Prioritize actions by economic impact and urgency

Continuous Optimization Requires Optimal Day-to-Day Decisions Optimize, either initial stock or stock adjustments (the orange area ) This recovers the difference between extremely inefficient solutions and an optimal solution – but the benefit decays over time as conditions changeIntroduce optimization into the supply support chain itself (the blue area ) A supply optimization system attempts to recover the remaining lost profit and operating margin, providing continuously optimal solutions Time Support Effectiveness Ultimately, the third step will be to eliminate periodic re-optimization in favor of “episodic” re-optimization

If you have questions or suggestions, please contact me:Robert Butler +1 831 649 3800 rab@tfdg.Com Thank you for your attention