Total productive maintenance (TPM) is a Japanese approach for involving production personnel in the maintenance process. The program is characterized by the 5-S foundation--this means 5 Japanese words (translated to 5 English words below) that describe a philosophy for doing maintenance work using teamwork and self directed work-forces.
Eliminate the unnecessary
The TPM program with English words noted above is described by Tokutaro Suzuki in TPM In
Process Industries, Productivity Press,
TPM is not as exciting as John Wayne ridding into town to attack a problem with his six-shooters blazing. However the results are longer lasting and more profitable. The carefully drilled team in the long run will always out-gun John Wayne!!!
The main questions are how much does TPM reduce failures (see below for
quantification of the savings) and how soon (the data shows approximately two
months and other programs have similar short time periods before results are
noticeable)? The details listed below were updated and
corrected from actual failure data submitted by Turbotech
1) Page down for this month's problem statement.
2) Return to the list of monthly problems by clicking here.
3) Bypass the background information and go directly to the problem statement by clicking here.
In a recent Turbomachinery Maintenance Congress in Brazil, a paper was given by Paulo S. De Oliviera, Jairo Torres, and Lucio A. Moreira Ivo of CEMAN--Central de Manutenção, Brazil. The paper was entitled: "Application Results of Nss, Minimum Flow, BEP and Operation Point Concepts in Evaluating Centrifugal Pumps Performance - Case Histories".
The paper showed declines in monthly maintenance interventions for a group of 114 ANSI pumps operating under various conditions in a monovinyl-chloride plant. Results of Figure 21 from the paper are shown in Figure 1 (along with audited data for failures).
The downward sloping trend line leads you to believe the maintenance department will be out of business in late 1996---this will only happen in the fantasies of naive people!
Note two special lines are marked on the trend chart. The first line in June '95 shows implementation of an action plan to make improvements based on a decision diagram. The second line is based on implementation of a TPM program in October '95.
At what point in time did the real improvements commence? Has TPM reduced the number of repairs?
The technical paper data has been updated with more information and correction (all failure data has deficiencies--audits of the data help improve the accuracy) of the date when the TPM program was instituted. Refer to the details shown below:
Earlier data reported the TPM program began on October 1995 rather than the correct date of August 1995. This date will change some of the trend lines--but does not change the conclusions. The summaries for each year paint a beautiful picture of reductions in failures--and the reductions occurred quickly after the program was started.
Figure 1 gives the appearance that all programs are simply continuation of favorable results which began before the reporting period. It appears that no identifiable program of significance has occurred. Figure 1 shows some scatter about the trend line and if the trend line is believable, then all is well by the end of 1996--a condition not expected by reasonable and experienced personnel.
Figure 2 takes another look at the data. The format of Figure 2 follows a reliability growth plot using the familiar Crow/AMSAA plots (formerly known as Duane/AMSAA plots). In this format cumulative failures are plotted against cumulative time. Stable processes produce straight lines and the slope of the line depends on improvements (the exponent is less than one), no changes (the trendline exponent is ~1), and deterioration is occurring and the exponent is greater than one.
Figure 2 shows two straight-line curves. The red line describes the "Before TPM" curve. The green line describes the "After TPM" curve where improvements are occurring faster than previously predicted.
The yellow zone between the two trend lines shows the number of failures avoided by use of TPM. In short, the program is quantified. After 36 months of data (and 29 months into the TPM effort) the trend lines allow computation of the saving using the equation N = ltb.
The Crow/AMSAA line slope b is very important information: 1) if b » 1 the system is not improving, 2) if b < 1 failures are occurring more slowly, and 3) if b > 1 then failures are occurring more quickly. As reliability engineers, our task is to put cusps on the trend lines so the lines have very shallow slopes! For more details see the November 2002 Problem Of The Month.
At the 36 month time, the predicted interventions without the TPM program is 34.65(36)0.947 = 1032 interventions. The predicted interventions with the TPM program is 77.49(36)0.529 = 516 interventions. This means 1032-516 = 516 interventions avoided by use of TPM. If the average cost of an intervention is US$1000, then 516interventions*US$1000/intervention = US$516000 saved in 28 months.
Note that the TPM effort began in June and was formally kicked-off in August (that’s only two months of priming the system so expenditures on the get ready effort cannot be too large).
Questions & Answers:
1) At what point in time did improvements commence?
Figure 2 shows modest (or no significant improvements) improvements in the trend line for the first 9 months (TPM was introduced at month number8). No break occurs in the trend line for the point at which an action plan for making improvements was introduced. A major break in the trend line occurs after the TPM program was introduced.
2) How many failures have been avoided?
Figure 3 shows a graph of expected failure versus actual failures. This information was derived from the statistics in Figure 2.
Notice the strong start with the TPM program. The TPM program had a fast drop in the number of interventions as they were cut by half very quickly--this occurred with the break-through of the TPM program. Now the theoretical failures have flattened-out, and to make a substantial improvement requires another break-through!
As a practical matter, we can project a monthly failure rate of 7 for 1998 and 6 for 1999. The practical objective is most likely to hold the program while attacking another area where costs are high. Thus a shift in emphasis will most likely be more beneficial than pressing for giant improvements.
3) What's the key event resulting in the improvements?
I had private conversations with Turbotech Engenharia engineers Paulo Oliveira and Jairo Torres during a recent trip to in Salvador, Bahia, Brazil l. Oliveira and Torres feel the key to improvement lies with the cooperative effort between production and maintenance to make improvements by educating and training the workforce using one-point or one-step lessons.
Oliveira and Torres attribute the one-step lessons to training operators and maintenance personnel to know the differences between right and wrong. Oliveria and Torres have over 300 one-step lessons in their book of instructions--each lesson is on one side of one sheet of paper in simple diagrams. One-step lessons are usually directed to correcting minor flaws and establishing basic equipment conditions. Look for a listing of one-step lessons at this site very soon.
The TPM approach works well with one-step lessons where few "rocket scientist" exist in the workforce. From the time schedule in Figure 3, the program is demonstrated to be quickly effective. In many cases, the TPM improvements are accomplished in the time it would take to "gear-up" for a reliability-centered maintenance (RCM) effort.
Many of the major causes behind frequent failures with pumps in Brazil stems from net specific speed problems (Nss). Nss is a subset of best efficiency point (BEP) problems which were quantified in the August '97 and September '97 problems of the month. Oliveira and Torres are attacking the Nss problem (when greater than 11,000--short life!!) by exception, whereas the TPM effort attacks many small problems in a variety of areas. You will find Turbotech Engenharia’s book [in Portugues] Bombas Centrífugas Passo a Passo [Centrifugal Pumps Step-by-Step] (1999) as a very practical instruction manual for operators and maintainers of centrifugal pumps. The book is simple, easy to understand, written to inform (not to impress you with mathematical formulas), the graphs, photos, and diagrams are worth the price of the book.
As Fred Geitner of Process
Machinery Engineering Services in
3) Can these TPM results be scaled directly for other applications?
Be careful in generalizing results of TPM to make specific cases apply to "all" applications. You need to know your specific details before you "guarantee" similar results. In general, when TPM is carefully applied, quick gains are achieved and you must "hold the course" for a long period to milk the gains from the system---don't look for "90 day wonders"--that only happens in the text books.
By the way, the reliability growth plots in Figure 2 were made using WinSMITH Visual software and the trend line equations resulted from the regression analysis.
Refer to the caveats on the Problem Of The Month Page about the limitations of the following solution. Maybe you have a better idea on how to solve the problem. Maybe you find where I've screwed-up the solution and you can point out my errors as you check my calculations. E-mail your comments, criticism, and corrections to: Paul Barringer by clicking here.
Technical tools are only interesting toys for engineers until results are converted into a business solution involving money and time. Complete your analysis with a bottom line which converts $'s and time so you have answers that will interest your management team!