HOW TO JUSTIFY MACHINERY IMPROVEMENTS USING RELIABILITY ENGINEERING PRINCIPLES by H. Paul Barringer Consultant Barringer & Associates, Inc. Humble, Texas and ‘Todd R. Monroe Principal Reliability Engineer Equistar Chemicals LP Deer Park, Texas Hi, Paul Barringer is a manufacturing engineering, and reliability consultant at Barringer & Associates, Inc., in Humble, Texas. He has more than 35 years of ‘engineering and manufacturing experience in design, production, quality, mainte- nance, and reliability of technical products. ‘Mr. Barringer is author of the basic reliability training course, “Reliability Engineering Principles” and is a contr ‘butor 10 The New Weibull Handbook, a reliability engineering text, published by Dr. Robert B. Abernethy. ‘Mr. Barringer has B.S. and M.S. degrees (Mechanical Engineering) from North Carolina State University, and participated in Harvard University’s three week Manufacturing Strategy conference. He has six U.S. patents and is a registered Professional Engineer in the State of Texas. Visit the world wide web site at Inip:/hwww:barringerl.com for other background details or send e-mail to hpaul @barringer!.com concerning LCC (or reliability issues, Todd R. Monroe isa Principal Reliability Engineer for Equistar Chemicals LP. in LaPorte, Texas, specializing in machinery reliability and mechanical seals for the complex. His present responsibilities include root cause failure analysis, reliability improvements for existing equipment, and specification, installation, and commissioning of new equipment and new process units. Prior to joining Equistar, he was with the Durametalic Corporation, working as an Applications Engineer Mr. Monroe has a B.S. degree (Mechanical Engineering) from Texas Tech University (1984). He is a registered Professional Engineer in the State of Texas, and a member of the Vibration Institute. ABSTRACT For the typical machinery engineer, the difficulties encountered with making reliability improvements le not with the “mechanics” of improvements, but with justifying the cost of improvements. It is difficult to translate sound engineering principles into terms that the accounting community can understand, m ‘The purpose of this paper is to equip machinery engineers with reliability engineering principles that translate “best practices” into ret present valve financial terms. Using the concepts of life-cycle costs, the effects of off-design conditions and poor installation practices will be shown to reduce the expected life of a pump by as ‘uch as 60 percent. A Monte Carlo simulation will be used t0 ‘compare commercial practices, good practices, and best practices for a specific pump installation, Life-cycle costs will be summarized, for each ease, in net present value, to select the best equipment choice LIFE-CYCLE COST INTRODUCTION Life-cycle costs (LCC) sum all total costs from inception to disposal for both equipment and projects (Barringer and Weber, 1996). The objective of LCC analysis is to choose the most cost- effective approach from a series of alternatives, so the least, long-term cost of ownershp is achieved. TLC is strongly influenced by equipment grade, the grade of installation and use practices, and maintenance strategies. The typical problem of specifying and justifying equipment is, producing the numbers and making them defensible inthe face of, ‘numerous conflicts: « Project engineers want fo minimize capital expenditures, ‘© Maintenance engineers want to minimize repair hours, ¢ Production wants to maximize uptime hours, «© Reliability engineers want to maximize equipment reliability, and ‘© Accounting wants to maximize net present value ofthe project ‘These conflicting issues, with no specific answers, result in a management edict to “buy cheap and complete the project quickly.” This paper will show how to find the numbers so an average engineer has a working tool to “buy right rather than buying cheap.” and finish che analysis quickly. LCC analysis helps engineers justify equipment and process selection based on total costs rather than inital purchase price. The sum of operation, maintenance, and disposal costs far exceed procurement costs. Lifecycle costs are total costs estimated to be Incurred in the design, development, production, operation, ‘maintenance, suppor, and final disposition of a major system over its anticipated useful life span (DOE, 1995). The best balance among cost elements is achieved when total LCC is minimized (Landers, 1996). As with most engineering tools, LCC provides best results when both engineering art and science are merged with good judgment. Procurement costs are widely used as the primary (and sometimes only) criteria for equipment or system selection—i.e.,m PROCEEDINGS OF THE 16TH INTERNATIONAL PUMP USERS SYMPOSIUM cheap is good. Procurement cost isa simple criterion. It is easy to use. I often results in bad financial decisions! Procurement costs tell only one part of the story. The major cost lies in the care and feeding of equipment during its life. Remember the adage attributed to John Ruston, “I’s unwise to pay too much, but it's foolish to spend too little” ‘Usually, procurement cost isthe only value in life-cycle cost that, is well known and clearly identified—but itis only the tp of the iceberg. Seeing the tip of an iceberg (similar to the obviousness of procurement cost) does not guarantee clear and safe passage around an iceberg. Hidden, underlying substructures of an iceberg (similar to the bulk of other costs associated with life-cycle costing for equipment and systems) contain the hazards. Life-cycle cost was conceived in the mid 1960s, and many original works on LCC are now out of print. Publications by Blanchard, et al. Blanchard and Fabrycky, 1990; Blanchard, 1992; Fabrycky and Bianchard, 1991), regarding life-cycle costs are now sources for a variety of LCC interests, LCC emphasizes business issues by enhancing economic compet itiveness to work for the lowest long-term cost of ownership. This requires engineers o worry about all cot detals—they must 1) sink like MBAs, and 2) act lke engineers for profit making enterprises. ches for acai a a ay [ooszmmdoss) —Cinmige cats ] Loum] Freeerwommm} [fermi ooo a ee cameo] LTrim] Lf aervconenn Figure 2, Acquisition Cost Tre. WHAT GOES INTO LIFE-CYCLE COSTS? avwcne Fecy wa Cos Denes ‘The basic cost tree for LCC starts simply. The tee has two bran- [tame —] [[eegcoma —] [fantom jon costs and sustaining costs, as shown in Figure I Tone cage “owng ase Lee] Hee) He |_prsramomee] |_f ma) mene Tome one — | gare TGS Tea See [Lites | [Laer ne Lp eames] [reer —] LD rm Somer ade wren Figure 1. Top Levels of Life-Cycle Cost Tree. Acquisition and sustaining costs are not mutually exclusive. IF you acquire equipment or processes, they always require extra ost to sustain the acquisition, and you cannot sustain without Someone having acquired the item. Acquisition and sustaining costs are found by gathering the corect inputs, building the input database, evaluating the LCC, and conducting sensitivity analyses to identiy cost drivers. In general, cost deals for the acquisition tree, shown in Figure 2, are usually identified and collected correctly. The collection of costs forthe sustaining tree shown in Figure isthe major problem! Frequently, the cost of sustaining equipment is two to 20 times the acquisition cost. The frst obvious cost (hardware acquisition) {is usually the smallest amount of cash that willbe spent during the life of the acquisition, and most sustaining expenses are not obvious. “The cost details must go into the corect ime buckets to make the life-cycle eos information useful. Cost duals ar important for ‘how much cost, and 2) when cost are incurred These details re used for calulating net present valve. Net present value is the single most important financial indicator for making decisions relating to capital issues For the sustaining tee, four items ae dificult to colect: + Replacementrenewal costs, ‘* Replacemenv/renewal transportation costs, + Suppordsupply maintenance costs, and © Operating costs Figure 3. Sustaining Cost Tree. Electrical costs are also particularly important because of high costs and varying loads on the equipment. ‘Most capital equipment authorizations ignore major portions of the sustaining cost tree as they lack details for the sustaining ‘expenditures. Based on some “justifications,” equipment never fails and, surprisingly, some of the equipment never uses clectcity!! When fullure costs are included, they appear as a percentage ofthe intial costs, and are spread evenly through every ‘year of the typical 20-year life fr the project. For wear-out failure ‘modes, the analysis & penalized by not including failures in the proper time span, Complications ariwe in the sustaining tree that are driven by planned costs in the acquisition tree. About 65 percent to 75 percent of the total LCC is set when the equipment is specified— and most decisions are based on the acquisition tree, which isthe smallest portion of the LCC! LCC is a process. The process provides for including appropriate costs, as shown in Figure 4, Appropriateness changes with each specific case as shown (Fabrycky and Blanchard, 1991, Appendix A). ENGINEERING FACTS LCC requires facts driven by data, Most engineers say they lack data, However, data are widely available as a starting point for LCC GBloch and Geitner, 1995). Often data reside in local ‘computer files, but it has not been analyzed or put to effective use, ‘Analysis can start with arithmetic and grow to more complicated Statistical analysis (Barringer and Weber, 1995), Follow the ‘Buidelines for each sep listed in Figure 4 to work out a typical engineering problem, Remember that a single right or wrongHOW TO JUSTIFY MACHINERY IMPROVEMENTS USING RELIABILITY ENGINEERING PRINCIPLES. 1 Figure 4, Process Flow for LCC Calculations. ‘method or solution does not exist—many methods and routes can bbe used to find LCC. If you disagree with the cost or life data, substitute your own values determined by local operating conditions, local costs, and local grades of equipment. Step 1: Define the Problem ‘The process requires installation of a new pump for a forthcoming modification. It is a 30 hp pump: 3 inch inlet * 1.5 inch outlet * 10 inch volute, with an 8 inch impeller. The pump provides water for transporting 30,000 Ib/hr of high-density polyethylene (HDPE) pellets. Ifthe pumping system fails, it shuts ‘down the process and penalizes the plant with gross margin losses ‘of $3000 per hour of failure. Step 2: Alternatives and Acquistions/Sustaining Costs Study two grades of pumps: ANSI, and ANSI enhanced with extra rigid baseplate. Note that API pumps are nota consideration, ‘because no physical need exists for utilization of the high grade features such as high temperature, etc. ‘Study two physical installation altemates: solo pump, and dual ‘pumps. The dual pumps will be run in sequence of one week online and one week offline. Study two maintenance strategies: fix when broken, and good ‘maintenance practices of replacing additional components when the pump has failed, so that reliability is improved and future failures delayed. ‘Study three grades of installation and use: commercial installation from quick/cheap installations and widely varying operating s Te condition, beter insalition and use practices with ceaonabe Tims impoted and best italation and bes we paces imposed @ Define the problem requiring LOC for precise instalation and small variations allowed during i pein ofthe equipment Of couse, cach pde of instalation i inresingly tore cpeaive to implement and contol —— |g Thestndy options and conditions sted above require solving 24 © | semen eptininig senna pe a wen Tile 1. Solve 24 Poblenso Find the Best Solution © | Prepue con beakdon smocureece | pacsaee Z eS @[— coeoseanaiyicalcotmaier fe a ae eee z [eciee © | catercost estates andeont models com — : a at © [Mae cos profes for each year of tly em aS 5 at t [om nev @ | Make breakeven chars for ahematives [rae mel ist sl el — al i Step 3: Prepare Cost Breakdown Structure/Tree © J Pareto chats of vital few cost contributors ach cas forthe cost breakdown structure wilincur os inthe t categories of Figue$ (pease rete io APPENDIX Tor Figur 5). ‘This is checlit for con to be collected Note that for this © | sensitive analysis of high cot and reasons analysis, the obvious costs that are generally defined correctly in thet projects Forte acqpsiion costs, such et project manageatct t costs, ele, are not included because they are common to all the Feedback| cases: Tis anlyis pays particular aneation fo he sstining cost @ | stay risks oF high cost items and occurences [PeSOGS] ses. This analysis ays part s i Capital costs forte acquisition tees are shown in Table 2. The cost in Table 2 ae for slo pump conigurtio, and al pomp ‘Select preferred course of action using LOC will double these acquisition costs. Replacement costs for the Susaining tree conceming each item are shown in Table 3. Time required for making replacements forthe sustaining tre is shown in Table 4. Time required for good maintenance practices are paced bby the item that fails, and assume all other replacements are accomplished within the alloted time for repair ofthe failure. Table 2. Capital Costs Detail Table 3. em Replacement Costs.1% PROCEEDINGS OF THE 16TH INTERNATIONAL PUMP USERS SYMPOSIUM Table 4. Replacement Times—Hours. Elapsed Repair Time (hours) ‘ANSI [ANSI Enhanced 7 13 Pump Bearin 7 Mechanical Seal 7 e 7 7 Motor Bearings 7 Replacement Motor] 7 ‘Motor Starter 1 System failure costs forthe sustaining tree concerning solo and dual pumps are shown in Table 5. Pump performance details and clecrical costs for the sustaining tree are summarized in Table 6 Details in Tables 2 through 6 provide cost information. They lack cost profiles in discrete time intervals for the sustaining cost tree ‘that will be used in NPV calculations. Table 5. Risk Costs for System Failure—Solo and Dual Pumps. Probability of simultaneous failure of ual pumps = Pp“ UAS where: Pp = probabilty of primary device failure in one year UAs = unavaliabilty ofthe standby device in one year Table 6. Pump Performance Details and Electrical Costs. [commerciat] oatar | Guat Practne | Paco | Practices feranponi fw 378° iors —[ too oom Pat Tie Sysa Reliability models are needed to find when end of component life occurs, as costly replacements follow death ofthe componeat. Details from the reliability models go into the sustaining costs. Every piece of equipment comprises components. Each ‘component has an inherent reliability, Inherent (or intrinsic) reliability is the best theoretical capability of components that can ‘be demonstrated ina laboratory environment Inherent component reliability is altered downward, as ‘measured by age-to-failure, by grade of installation, and how the ‘equipment is used. Gride is a rank indication of the degree of refinement, features, or capabilities for installation and operation. ‘The grades of equipment installation/use practices (and thus the costs) inthe acquisition tree are precursors of failure costs covered in the sustaining tree. LOC accuracy improves and benefits from ‘quantification of grade issues, just as LCC analysis improves by use of probabilistic aralysis rather than deterministic analysis. Most engineers know that nothing dies on schedule and nothing ‘gets repaired on schedule—the problem is how to quantify the details ‘Very high-grade instillations and very high-grade practices for ‘operation of the equipment demonstrate (by long ages-to-falure) most ofthe inherent equipment life is usable. However, low-grade installations and low-grade operations destroy inherent life of equipment. The key for LCC calculations is accurate reflection of effects of practices on inherent reliability. This is frequently accomplished by laboratory tests on low-cost electronic ‘components, but seldcm quantified on high-priced mechanical ‘components’ Thus, the effects of practices on inherent life must be ‘obtained by surveys from experts in the field, Field surveys may not be the desired method of acquiring data, but itis the most practical method considering time and costs, Ths forms the art of. engineering that is used with the science of engineering to derive practical answers in real time. Engineering drawings specify the grade of installation (and frequently they speci'y the grade of operation). Often, the production department specifies the grade of operation. The grade ‘of equipment installation and operation needs to be priced-out as part of the specification process. It cannot be ignored—it affects life-cycle costs. Maintenance departments and operation departments, if total productive maintenance (TPM) oriented, specify the grade of ‘maintenance. Most engineers, during their carer, are accused of ‘gold plating over engineering, and wasting money on the grade of installation. Engineers are aware of how the installation grade strongly influences the number of failures for the sustaining tre, but lack details for quantifying their opinions. ‘Good LCC analysis responds to accusations of detractors by providing monetary rests to refute charges of gold plating, over esign, high cost, and other overstatementsexaggerations. The resulting life multipliers, from Table 7, alter the inherent reliability of equipment and change base failure rates (lower failure rates) 0 predict actual failure res (higher failure rates). ‘The multipliers from Table 7 are used in Table & (please refer to APPENDIX for Table 8) to alter the inherent reliability model. Notice how the ANSI model and the ANSI enhanced models have different inherent life. For the best practices, only thee percent of | the inherent life is lost, compared with 44 percent loss of life for better practices, and 97 percent loss of inherent life for the lowest ‘grade commercial practices. ‘Table 8 also shows taxonomy (the classification of equipment in ‘an ordered system that indicates natural relationships) effects on life. Life of the mechanical portion of the system is different from the electrical system, When both are considered together, system life is much shorter, The taxonomy definitions demonstrate why different organizations talk about different numbers for the same device. Reciprocals of the mean time to failure (MTTF), in Table 8, give failure rates, which could be used for deterministic computations of net present value (NPV). Deterministic NPV calculations are simpler, but often give inaccurate results when ‘components have wear-out failure modes as shown above, ‘Table 8 uses Weibull statistics to characterize component life (Abernethy, 1998). Weibull descriptors are tools of choice for reliability engineers. Beta value implies failure modes for the ‘components. In Table 8, the beta values are greater than unity and imply wear-out failure modes.HOW TO JUSTIFY MACHINERY IMPROVEMENTS USING RELIABILITY ENGINEERING PRINCIPLES us Table 7. Centrifugal Pump Practices, Life, and Costs. ‘The cost data above must also include maintenance replacement strategies, as shown in Table 9, With the input data from Tables 2 through 9, the search can begin for minimizing the lowest long term cost of ownership as found by LCC. Every LCC example has its own unique set of costs and problems to solve for minimizing LCC. Remember that minimizing LCC pushes up NPV and creates wealth for stockholders, Finding LCC requires finding details for ‘both acquisition and sustaining costs with many details. The most [3 | «| = | «| 7 | s |] | ler i lass earl — opto | — amr — aes — ser} — oe lms fie ti nb lowers Ts Tals — rah — apt Ta a sa masta west meet eat Test ae ro Peng PT et tote ntl det ee tat" alte vel fret — a [fra 0 att —a Bont — Bien Bisolt zits] Saco Fc Ra on fe on oe ee Te | os [reer Ca EO exer eomle te Jr vaue [Fe evar Je-Fia Ras Or Rens Geaures aa oo poslive nara an fe wgivo urban on 3% ae 8 ISN bases ras =HOW TO JUSTIFY MACHINERY IMPROVEMENTS USING RELIABILITY ENGINEERING PRINCIPLES 13 Table 13. Top 10 Alternatives. TopTen Alternative From Monte Carlo Simulation caro [Itaaton & | Waoterance ve rc 7 ANS Favor sonar TES [TY 5 2 ‘ANS Gator —{ Fa ner Broken [S270 [§lor-ont wae 88% 3 ANSI Best Good Paces [$0368 [s (os 225] ere 7% | 45%, 4 "ANSI Batter ‘Good Practices | $27,458 | $(68,913)] 4224 “7% [1.0% 5 NSIS Bottor | Fx When Broken_[ $95.496 |$ (72.611)[ 8%. "72% | 6.1%. 6 ANSI Batter ‘Good Practices | $35,436 [§ (73,868)] 60% 8.1% [2.9% 7 ANSI Best | Fix When Broken | $39,086 [§ (r4253]| 75% “9.6% | 212% @ ANSI Bost ‘Good Practoos | $39,046 | $ 74.96a)| 75% | ~to.7% | aa% 2 ANSI: Best__| Fix When Broken_| § 19.528 | $ a2,685)| 75% | -220% | 21.1%. 10 “ANSI Best ‘Good Practoes [$18.58] § 62,722) 75% | 20.2% | 22.4% “Altornative. ae Trstallation & Rank, $20_| eo prectee 7 ANT fForWhen & oven | $3008” [ S67 729) = = 2 ANSI {us| Bate | Fi Wren roten[§ 2s [sersort tex ——[ ase aS [ani [Dual Best | Goa Practoes | $0,258 [$(68 225 45% 4 ‘ANSI | Dual | Better | Good Pracicas_| $27.68 | sga.ar9 EK [ANSI Dual | Bate Fic When Broxen| s 25.05 [Stra 81% [ANSI [Dual Batter | Good Practon_| $35.00} (70.064 29% 7 [ANSI [ Dual Set [Penn Sen $e ares 212% ‘3 [-aNsis [Dual [Best] Goos Practoss | $50 : 222% ‘9 —[-Aanst-[ So [Bat Fix When eroxon | s1500_| Sees 21.1% ww —Lansi-- [sis [pest —[Gouiracte Tans Ispeaiel res hoes Lez Force To ($67,728) And Find Required Cost Reduction Trade-down REFERENCES Blanchard, B.S. and Fabrycky, W. J, 1990, Systems Engineering Abemethy, R. B., 1998, The New Weibull Handbook, Third Edition, Houston, Texas: Gulf Publishing Company. Barringer, H. P. and Weber, D. P, 1995, “Where's My Data for Making Reliability Improvements,” Fourth International Conference on Process Plant Reliability, Houston, Texas: Gulf Publishing Company. ‘Barringer, H. P_and Weber, D.P, 1996, “Life Cycle Cost Tutorial” Fifth International Conference on Process Plant Reliability, Houston, Texas: Gulf Publishing Company. Barringer, H. P, 1998, “Download Free Life-Cycle Cost Software,” hup:/lwww barringer|.com Bloch, H. P. and Geitner,F.K., 1995, “Simplified Life-Cycle Cost ‘Computations Applied in the Hydrocarbon Processing Industries,” Fourth Intemational Conference on Process Plant Reliability, Houston, Texas: Gulf Publishing Company. Brennan, J. R., Strcener, J. T, Huff, H. H., and Burton, S. A. 1985, “Reliability, Life Cycle Costs (LCC) and Warranty’ Lecture notes from a General Electric inhouse tutorial Blanchard, B.S., 1991, “Design to Cost, Life-Cycle Cost” Tutorial notes, Annual Reliability and Maintanability Symposium, available from Evans Associates, Durham, North Carolina Blanchard, B.S. 1992, Logistics Engineering and Management, Fourth Edition, Englewood Cliffs, New Jersey: Prentice-Hall Blanchard, B. S., Verma, D., and Peterson, E. L., 1995, ‘Maintainability: A Key 10 Effective Serviceability and ‘Maintenance Management, Englewood Cliffs, New Jersey: Prentice-Hall. ‘and Analysis, Second Edition, Englewood Cliffs, New Jersey Prentice-Hall Department of Energy (DOE), 1995, hup:/vww.em.doe gov ‘Afeabb/ovpstpife hin, posted 4/12/1995 on the world wide web, Fabrycky, W. J. and Blanchard, B. S., 1991, Life-Cycle Cost and Economic Analysis, Englewood Ciifls, New Jersey: Prentice- Hall Followell, D. A., 1995, “Enhancing Supportability Through Life- Cycle Definitions," 1995 Proceedings Annual Reliability and Maintainability Symposium, available from Evans Associates, Durham, North Carol. Kececioglu, D., 1995, Mainainabilin, Availability, & Operational Readiness Engineering, Upper Saddle River, New Jersey: Prentice Hall PTR. Landers, R. R., 1996, Product Assurance Dictionary, Marlton, New Jersey: Marlton Publishers. Pecht, M., 1995, Product Reliability, Maintainability, and Supportability Handbook, New York, New York: CRC Press, Raheja, D. G., 1991, Assurance Technologies, New York, New York: MeGraw-Hill, In. ‘Weisz, J., 1996, “An Integrated Approach to Optimizing System Cost Effectiveness.” 1996 Tutorial notes, Annual Reliability and Maintainability Symposium, available from Evans Associates, Durham, North Carolina ‘Yates, W. D., 1995, “Design Simulation Tool to Improve Product Reliability,” 1995 Proceedings Annual Reliability and186 PROCEEDINGS OF THE 16TH INTERNATIONAL PUMP USERS SYMPOSIUM Maintainability Symposium, available from Evans Associates, Durham, North Carolina. BIBLIOGRAPHY ‘Cumulative Indexes,” 1996 Proceedings Annual Reliability and ‘Maintainability Symposium, availabe from Evans Associates, Durham, North Carolina, page ex-29 for LCC references.