Proactive maintenance and reactive repair
Ernst W. Diehl
Massachusetts Institute of Technology
Center for Organizational Learning
Room E60-375
30 Memorial Drive
Cambridge, MA 02142-1347
“E-Mail: ewdiehl @mit.edu
ABSTRACT
Each month a major US telephone company receives 200 000 calls from customers who
have a problem with their telephone service. The company places a high emphasis on
reducing the repair cost caused by the high volume of complaints. Using a simulation
model, the company wants to understand how more proactive maintenance can reduce the
need for repair calls.
The simulation model presented in this paper reveals that within proactive maintenance we
need to distinguish between at least 3 different policy levers: (1) Discover the problem
before the customer notices it; (2) Do the repair with such a quality that you do not have to
repair the same problem twice.; (3) Make your physical plant more reliable. Each of the 3
policies will have different cost savings and different payoff delays. The simulation model
allows the company to allocate investments in each of the 3 areas and to test which
investment mix fits the overall company objectives best.
INTRODUCTION
Imagine you are the head of maintenance for a large regional phone carrier. You're in
charge of the repair and maintenance of over 14 million phone lines, of which 220
thousand have trouble each month. Headquarters has given you a mandate to reduce that
figure by 20% over the next 3 years. To accomplish this, you have assembled a cross-
functional team which has met several times. So far the team has generated a whole list of
counter measures, such as providing more training to your repair technicians or investing in
more reliable cable technology and you have had some heated discussions about the relative
benefit of each one. But other than conflicting data and anecdotal evidence, you have no
way of testing the relative benefits of each measure, or of seeing what combination of
actions will have the greatest benefit.
What you need is a tool that will give you some way to look at the various maintenance
strategies in a cohesive way. To assist in this process, your MIS department has given you
anew computer simulator, which enables you to experiment with alternative scenarios and
see the output in reports that are identical to the maintenance reports you view on a monthly
basis. You call your team together for an afternoon meeting, in which you plan to use the
simulator to see the relative outcomes of each strategy.
BACKGROUND
The author found himself placed in a similar context as the one described above when he
was asked to help in the design of such a computer simulation. One of the most important
learning occurred during the development of Figure 1 that breaks down the 14.4 million
lines into four main categories: lines that don't have a problem; “re-active repairs,” which
\o5
have been initiated by customer complaints; “‘pro-active repairs,” which are the problems
the company discovered before the customer did; and those lines that are not showing
problems yet, but are anticipated to cause problems within the next three months.
Figure 1: Overview screen
S: Investment
(000 $)
Quality (%) : 80 _ 200
Discovery (%) “102° ~~ 100 ~
Reliability (/1000/mo) 13.35 400
Defects (/1000/mo) 13.35 N/A
Trouble Status (in 1000 Lines)
No Undiscovered Known
Trouble Trouble Trouble
Defects Discovery
(per 1000 per month) 02
13.
35 Trouble Status
installed Base
Customers
44 1
Quality
80 Reactive
Proactive
Repairs
Rather than continue thinking in terms of one long list of counter-measures, the
development team began to see each item on the list as falling into one of three distinct
categories:
* Discovery. Currently, the company only discovers 2% of all problems, while the
customers uncover 98% of all problems. Investing in discovery means taking measures to
increase the number of proactive repairs that the company makes, in order to reduce the
number of problems that customers call in.
* Quality. Past data has shown that, of the 220 thousand lines repaired each month,
20 percent of them were not fixed properly and will need service again in three months.
(ab
Efforts such as investing in training and creating more standardized repair procedures could
boost quality and reduce the need for future repairs.
¢ Reliability. A number of factors in the initial installation of phone lines, such as the
materials used and the location of the lines, that affect the expected failure rate of the phone
lines. By investing in reliability, you can prevent problems before they occur.
COMPARING INVESTMENT STRATEGIES
All of those counter measures are not free, however:~ Your current maintenance budget is
$700,000 per month, of which you are currently spending $100,000 on discovery
(proactive maintenance), $200,000 on quality, $400,000 on that reliability. Management
has allocated $300,000 more maintenance budget over the next several years. The question
facing you and your team is, of the three categories outlined above, where should you
allocate your money in order to maximize the savings gained by the investments.
The simulator contains an interface that provides a comparison with the current "base case"
strategy. Figures 2-5 depict a scenario where we invest $1,000,000 in quality and nothing
in discovery or reliability.
S: cost VOLUME UNITCOST
— Monthly Report } (ons) {o09) ‘s)
Trouble Repair
Proactive 0 ie) 20
Reactive 14114 235 60
Investments
Trouble Discovery 0
Repair Quality 1000 Month
Reliability ie) 50
Total 15114
Figure 3: Cost comparison screen
Cost Comparison (000 $)
Month: 50
Current BaseCase Difference
= Monthly
Lost Revenue
Lost Customers 765 1132 -367
Lost Access Time 396 370 26
Maintenance Cost
Repair Cost 14114 13260 854
Investments 1000 700 300
Total 16275 15461 813
VO
Figure 4: Cost per month comparison Figure 5: Accumulated comparison
10005 153007
89754
26507
-3675 4
-1000 4, r + -10000 +, 7 7
0 30 60 fo) 30 60
O Total Cost A (Current-Base) Lost Access Time A (Current-Base)
© Lost Customer A (Current-Base) * Repair Cost A (Current-Base) ++ Investments A
Comparing scenarios reveals that not all maintenance efforts are created equal. Figure 6
compares 3 different investment strategies to the base case. In each case we have added
$400K to the monthly investments made in either quality, discovery, or reliability.
‘igure 6: C iso1 inst
Month when Total payoff Payoff in Repairs (% BC) Lost
total payoff>0 in month60 month60 in month 60 Customers
Base Case N/A 0 0 100.0 1331
BC + $300K for Quality 19 17141 536 95.4 1024
BC + $300K for Reliability 36 21439 1213 88.6 1259
BC + $300K for Discovery 10 19595 441 100.0 1232
Investment in reliability provides us with the biggest payoff ($21,439,000) and the largest
on-going benefits ($1,213,000). At month 60, repair volume has shrunk to 88.6% of the
base case and continues to fall. Under the current assumptions, investment in discovery has
the quickest payoff. It reduces cost, since it is assumed that it is less expensive to repair a
mistake if you schedule your repairs in advance. However, the total number of repairs to be
made remain constant. The simulator makes the assumption that some customers leave us
for a competitor if they have to endure the same repair twice or more often within 3
months. A investment in quality reduces the repeat repairs and helps us retain the most
customers.
Understanding the systemic differences of investments in quality, reliability and discovery
allows to design a mix of countermeasures that result in a short-term payoff and provide for
a long-term fundamental reduction in repairs necessary.
\ah
