The volume of outputfrom the first live computer run of master scheduling and MRP may be so great that the users can’t handle it all.. If master scheduling and MRP don’t work properly dur
Trang 1the point of implementing ERP if it’s just going to deliver the samelousy information that the current system provides?
That’s the problem with the parallel approach for ERP’s planningand scheduling tools
Big Bang
The inability to do a parallel leads some people to jump way over to
the other side of the fence, and do what’s called a big-bang cutover.
We call it “you bet your company,” and we recommend against it orously and without reservation
vig-Here’s an example of a big-bang implementation, as explained by
an unenlightened project leader:
We’ve got master scheduling and MRP all programmed, tested,and debugged We’re going to run it live over the weekend On Fri-day afternoon, we’re going to back a pickup truck into the pro-duction control office and the computer room, throw all theprograms, disks, tapes, procedures, forms, and so forth into thetruck and take ’em down to the incinerator
This gives new meaning to the phrase “burning one’s bridges.” Abig-bang cutover (also called “cold turkey”) carries with it two prob-lems, the first one being that ERP may fail The volume of outputfrom the first live computer run of master scheduling and MRP may
be so great that the users can’t handle it all By the time they workthrough about a quarter of that output, a week’s gone by and thenwhat happens? Master scheduling and MRP are run again, and herecomes another big pile of output The result: The users are inundatedand your ERP effort has failed
Folks, that’s the least of it The second problem is far more severe:You may lose your ability to ship product Some companies who’vedone a big bang have lost their ability to order material and releaseproduction The old system can’t help them because they stoppedrunning it some weeks ago, and the data isn’t current ERP isn’t help-ing them; it’s overwhelming them
By the time they realize the seriousness of the problem, they oftencan’t go back to the old system because the inventory balances and
222 ERP: M I H
Trang 2other data aren’t valid any longer, and it might be a nearly impossiblejob to reconstruct it.2
A company that can’t order material and release production willsooner or later lose its ability to ship product A company that can’tship product will, sooner or later, go out of business
Some organizations get lucky and muddle through without greatdifficulty In other cases, it’s far more serious Although we’re notaware of any company that has actually gone out of business for thisreason, there are some who’ve come close The people we knowwho’ve lived through a cold-turkey cutover never want to do it again
Never Don’t do it.
The Pilot Approach
The right way to do it is with a pilot Select a group of products, or
one product, or a part of one product, which involve no more thanseveral hundred part numbers in all—and do a big bang on those.The purpose is to prove that master scheduling and MRP are work-ing before cutting over all 5,000 or 50,000 or 500,000 items onto thesystem The phrase MS/MRP working refers to two things: the tech-nical side (does the software work properly?) and the users’ side (dothe people understand and believe what it’s telling them, and do theyknow what to do?)
If master scheduling and MRP don’t work properly during the lot, it’s not a major problem Almost all the items are still being or-dered via the old system, except for the few hundred in the pilot.These can be handled by putting them back on the old system or per-haps doing them manually What’s also necessary is to focus on whymaster scheduling and MRP aren’t working properly and fix it Thepeople have the time to do that if they’re not being inundated withoutput on 5,000 or 50,000 or 500,000 items
pi-What do we mean when we ask: “Is it working?” Simply, is it dicting the shortages? Is it generating correct recommendations torelease orders, and to reschedule orders in and out? Does the master
pre-2 Even those cases when they can go back to the old system are very difficult Why? Because they tried to implement ERP, and it didn’t work Now they’re back to run- ning the old system, not ERP, and they’ll have to decide what to do now, where to
go from here A most unfortunate situation.
Trang 3schedule for the pilot product(s) reflect what’s actually being made?Can customer orders be promised with confidence using the avail-able-to-promise function within master scheduling? Answering yes
to those kinds of questions means it’s working
By the way, while we’re talking about pilots, we should point outthat it’s generally a good idea to pilot as many processes as possible
So far in this book, we’ve talked about piloting Sales & OperationsPlanning, master scheduling, and MRP Where practical, you shouldalso pilot sales forecasting, inventory transaction processing, andbill of material creation and maintenance Look for opportunities topilot Avoid big bangs—or even little bangs
THREEKINDS OFPILOTS
Doing it right means using three different types of pilots—the puter pilot, the conference room pilot, and the live pilot
com-1 The computer pilot.
This simply means checking out the hardware and software verythoroughly It means running the programs on the computer, and de-bugging them (Yes, there will be bugs in the software no matter howmuch you paid for it or how large the customer base.) This processshould begin as soon as the programs are available
Often, the computer pilot will deal initially with dummy items anddummy data This should come with the software package Thedummy products are things like bicycles, pen and pencil sets, and thedummy data are transactions made up to test the programs Then, ifpractical, run the new programs using real data from the company,using as much data as can readily be put into the system
Next, do volume testing Sooner or later, you’ll need a program tocopy your current data into the new formats required by the newsoftware Get that program sooner Then copy your files over, and dovolume testing You’re looking for problems with run times, storage,degradation of response times, whatever Who knows, your com-puter may need more speed, more storage, more of both (Doingone’s homework at the onset of the project means recognizing thesepossibilities, and putting contingency money into the project budget
to cover them if needed.)
224 ERP: M I H
Trang 4In addition to hardware, other major objectives of the computer lot are to ensure the software works on the computer and to learnmore about it The key players are the systems and data processingstaff, usually with some help from one or more project team members.
pi-2 The conference room pilot.
This follows the computer pilot The main objectives of the ference room pilot are education and training: for the users to learnmore about the software, to learn how to use it, and to manage theirpart of the business with it and make sure it fits the business Thisprocess can also help to establish procedures and to identify areasthat may require policy directions
con-The key people involved are the users, primarily the folks in tomer service, the master scheduler(s), the material planners, andprobably some people from purchasing They meet three to fivetimes per week in a conference room equipped with at least one workstation for every two people The items involved are real-world items,normally ones that will be involved in the live pilot The data, how-ever, will be dummy data, for two reasons:
cus-a Live data shouldn’t be used because the company’s still notready to run this thing for real Everyone’s still in the learningand testing mode
b It’s important to exercise the total system (people, as well assoftware) as much as possible Some of the dummy data forthe conference room pilot should be created so it will present
as many challenges as possible to the people and the software.One technique that works nicely is for a key person, perhaps the proj-ect leader, to play Murphy (as in Murphy’s Law—“whatever can gowrong, will go wrong”) As the conference room pilot is being oper-ated, Murphy periodically appears and scraps out an entire lot ofproduction or becomes a supplier who’ll be three weeks late shipping
or causes a machine to go down or generates a mandatory and mediate engineering change.3
im-3 We can remember one project leader who had a sweatshirt made up It was navy blue with MURPHY printed in red gothic lettering.
Trang 5Murphy needs to determine if the players know the right responses
to both major pieces of the system:
a The computer side—the hardware and the software Do thepeople know how to enter and promise customer orders, how toupdate the master schedule, how to use pegging, firm plannedorders, and the other technical functions within the software?
b The people part—and this gets at feedback Do the plannersknow whom to give feedback to if they can’t solve an avail-ability problem on one of their items? Does the master sched-uler know how and whom to notify in Customer Service if acustomer order is going to be missed? Do the Customer Ser-vice people know to notify the customer as soon as they know
a shipment will be missed?
This last point gets at the important ERP principle of “silence isapproval.” This refers to mandatory feedback when something goeswrong In a Class A company, feedback is part of everyone’s job:sales, planning, plant, purchasing, and suppliers As long as thingsare going well, no one needs to say anything However, as soon aspeople become aware that one of their schedules won’t be met, it’stheir job to provide immediate feedback to their customer, whichcould be the next work center, the real end customer, or someone inbetween
Presenting the people and the software with difficult challenges inthe conference room pilot will pay dividends in the live pilot and cut-over stages One company we worked with used a slogan during theirbusiness meetings and conference room pilot: Make It Fail Super!This is another version of bulletproofing During the conferenceroom phase, they worked hard at exposing the weak spots, findingthe problems, making it fail The reason: so that in the live pilot
phase it would work.4
The conference room pilot should be run until the users reallyknow the system Here are some good tests for readiness:
226 ERP: M I H
4 One of your authors used to fly airplanes for a living, and crashed many times Fortunately, it was always in the simulator The fact that he never crashed in the real world is due in large part to the simulator The conference room pilot is like a simu- lator; it allows us to crash but without serious consequences.
Trang 6a Ask the users, before they enter a transaction into the system,what the results of that transaction will be When they canroutinely predict what will result, they know the system well.
b Select several master schedule and MRP output reports (orscreens) at random Ask the users to explain what every num-ber on the page means, why it’s there, how it got there, and soforth When they can do that routinely, they’ve got a goodgrasp of what’s going on
c Are the people talking to each other? Are the feedback ages in place? Do the people know to whom they owe feed-back when things go wrong? The essence of successful ERP ispeople communicating with each other Remember, this is a
link-people system, made possible by the computer.
If the prior steps have been done correctly and the supporting ments are in place, the conference room pilot should not take morethan a month or so
ele-Jerry Clement of the Oliver Wight group has a fine approach todealing with this issue: “After the conference room pilot is virtuallycomplete, I recommend running a full business simulation with boththe outside and inside experts totally hands-off If everyone executeswith comfort, you’re ready to install I go one step further: I give theend users a veto over going live If they don’t feel we’re ready, we must
fix the issue before we go live Think about hearing the end users ing, ‘Hey, this stuff really works What are we waiting for? Let’s turnthis baby on!’ I hear that after a good conference room pilot when theend users really believed they counted.”
say-3 The live pilot.
This is that moment of truth we mentioned earlier It’s when ter scheduling and Material Requirements Planning go into opera-tion for the first time in the real world The objective of the live pilot
mas-is to prove master scheduling and MRP will work within the
com-pany Until then, that can’t be said All that one could say up to thatpoint are things like: “It should work,” “We think it’ll work,” “It re-ally ought to work.” Only after the live pilot has been run successfullycan the people say, “It works.”
Trang 7Before we get into the details of the live pilot, let’s recap what we’vecovered so far by taking a look at Figure 11-2.
Selecting the Live Pilot
What are the criteria for a good live pilot? Some of the considerationsare:
Computer
ConferenceRoom
Live
Systems peopleProject leader
Customer svc
(Order entry)Master sched’r
MRP plannersSystems analystProject leader
Customer svc
(Order entry)Master sched’r
MRP plannersProject leader
Dummy/dummyLive/dummyLive/live
1 Learn more aboutthe software
2 Discover bugs
3 Check for problemswith run time,response time, andstorage
1 Do further usereducation andtraining
2 Build in feedback
3 Verify that thesoftware fits thebusiness
1 Use the system inthe real world
2 Prove that it works
3 Obtain a sign-offfrom the users
Team-Fly®
Trang 8whelmed Try to keep the total number of items (products, nents, raw materials) to less than 500.
compo-2 Product orientation.
The pilot should represent all the items for an entire product ily (in the case of a simple product, such as clothing or cosmetics), asingle product (moderately complex products, like some office equip-ment), or a part of a product (highly complex products, such as air-craft or machine tools) In the last example, the pilot might be oneleg in the bill of materials or a modular planning bill for an option
fam-3 Good cross-section.
The pilot group should contain a good mix of finished products,subassemblies or intermediates, manufactured items, purchaseditems, and raw materials
4 Relatively self-contained.
The fewer common parts contained in the pilot, the better Itemsused in both the pilot product and others will not give a good test ofMRP MRP will not be aware of all the requirements for those items.The usual way of handling this is to post the MRP-generated re-quirements back to the old system Some degree of commonality isalmost always present (raw materials, in many cases), but try to pick
a pilot where it’s at a minimum
5 Best planner.
If the company has material planners already and they’re ized on a product basis, try to run the pilot on the product handled
organ-by the best planner This is a people-intensive process, and it needs
to have as much going for it as possible
Look Before You Leap
Let’s consider what has to be in place prior to the live pilot One ment is a successful conference room pilot, where the users have
Trang 9proven they understand the system thoroughly The other key ments are data integrity, education, and training Please refer to theImplementers’ Checklist at the end of this chapter.
ele-The project team should address the first six entries on the list All must be answered yes The project leader then reports the re-sults to the executive steering committee and asks for formalpermission to launch the live pilot Only after that’s received shouldthey proceed
check-Operating the Live Pilot
When everything’s in place and ready to go, start running the pilotitems on MS/MRP and stop running them on the old system Theobjectives are to prove MS/MRP is working and to obtain user sign-off Is it predicting the shortages, giving correct recommendations,and so forth? Are the users, the master scheduler(s), and the materialplanners making the proper responses and taking the correct action?Are the users prepared to state formally that they can run their part
of the business with these tools? If the users are unwilling and/or able to sign off on the system, then one of several factors is probablypresent:
un-•It’s not working properly
•They don’t understand it
•Both of the above
In any of these cases, the very worst thing would be to proceed intothe cutover phase—to put all the remaining items onto the new sys-tem First, aggressively go after the problem: Either fix the elementthat’s not working properly or correct the deficiency in educationand training that’s causing the user not to understand it, or both.Run the live pilot for as long as it takes
Don’t go beyond the pilot stage until it’s working and the usershave signed off This is one area where the aggressive implementa-tion mentality must take a back seat Everyone—executive steeringcommittee, project team, users—should understand that the com-pany won’t go beyond the live pilot until it’s been proven to work anduntil the users are comfortable with it
230 ERP: M I H
Trang 10Plan to run the live pilot for about a month, or longer if turing cycles are long and speeds are slow It’s essential to observehow the man/machine system performs over a number of weeks toprove it really works A week is not enough time, and a quarter isprobably too long (for planning purposes) for most companies.During the live pilot, don’t neglect training Get the other plan-ning people as close to the pilot operation as possible, without get-ting in the way of the folks who are operating it People not involved
manufac-in the pilot need all the manufac-input they can get because they’ll be on thefiring line soon when the rest of the items are cut over to ERP.The pilot must be successful to go to cutover, and visible successsupports behavior change Therefore, make sure the other planningfolks can see the success
CUTOVER
Once the live pilot is working well and the users are comfortable with
it, it’s time to cut over the rest of the items onto the master scheduleand MRP
There are two different ways to cut over It can be done in onegroup, or the remaining items can be divided into multiple groupsand cut over one at a time
The multiple-group approach is preferable because it has the lowing advantages:
fol-1 It’s less risky It represents a more controlled process.
2 It’s easier on the people If the first group to cut over belongs to
planner A, then planners B and C should be deeply involved
in helping planner A It reduces planner A’s workload, andalso provides additional training for B and C When planner
B cuts over, planners A and C can help him or her
The multiple-group approach, on the other hand, may not always
be practical and/or necessary In some companies, there is so muchcommonality of components that it’s difficult to isolate groups Thismeans that during the cutover process, many items will be partiallybut not totally on the new system The difficulties in passing re-quirements from the new system to the old (or vice versa) can easily
Trang 11outweigh the benefits gained from using multiple groups In thiscase, the one-group approach would probably be best It’s usuallybetter at this point to move ahead quickly rather than to spend lots
of time and effort transferring requirements for common parts.Sometimes the multiple-group approach may not be necessary Acompany with only a few thousand items or less may conclude theirentire population of items is small enough, and there’s no real need
to break it down any further
A WORDABOUTTIMING
Sometimes companies get hung up on a timing issue with cutover,specifically an accounting cutoff Don’t delay cutover for any appre-ciable amount of time—waiting for the beginning of the month, thebeginning of the quarter, or (shudder) the new fiscal year Rather, thesystems folks should have any necessary bridging programs ready to
go to feed data from the master schedule and MRP side of ERP intothe accounting systems In this way, cutover can occur as soon aspractical and not be delayed waiting for the passage of time
THENEED FORFEEDBACK DURINGCUTOVER
There’s a potential dilemma here for you job shop people:
•This cutover is a phase I activity The master scheduling andMRP planning tools are being made operational
•However, closing the loop in a job shop is a phase II activity, whichcomes later (Even in a flow shop, where plant scheduling can bemade operational in phase I, supplier scheduling won’t be fully im-plemented until phase II, for reasons we’ll cover later.) How canmaster scheduling and MRP be made operational prior to that?The answer is that there must be a form of loop closing even dur-ing phase I It’s essential Without feedback from the plant and pur-chasing, the planning people won’t be notified when jobs won’t becompleted on schedule They must have that feedback, or they willnot be able to keep the order dates valid Therefore, the principle ofsilence is approval applies here, even at this early stage This means
232 ERP: M I H
Trang 12that anticipated delay reporting from both the plant and purchasingmust be implemented as part of phase 1 However, there’s even a bitmore to it than that for you job shop folks.
At this point, the company’s beginning to operate with the formalpriority planning system (MS/MRP) but doesn’t yet have the prior-ity execution system (the dispatching portion of plant floor control)
in place Given good feedback, order due dates can be kept valid and
up to date, but the tools to communicate those changing priorities tothe plant floor still aren’t available Further, without Capacity Re-quirements Planning, there’s no specific, detailed visibility into fu-ture overloads and underloads at all the work centers on the plantfloor
A good approach here is to develop an interim, simple, possiblycrude, plant scheduling system Use it to get the job done until thefull-blown plant floor control system is on the air This interim sys-tem is usually manual, not computerized, and operates with orderdue dates and possibly a simplified back scheduling approach (Ex-ample 1: Job A has a four-week lead time It’s due two weeks fromnow It should be 50 percent finished Is it 50 percent finished? If not,
it should be given priority Example 2: Back schedule from the orderdue date assuming all operations take the same amount of time Setoperation due dates accordingly.)
In addition, it’s highly desirable to assign one or more plant peoplefull-time to the project during this transition phase This person’s re-sponsibility is to help the folks on the plant floor work on the rightjobs He or she maintains close contact with the interim plant sched-uling system, with the material planners, and with the foremen Shefinds out about the reschedules coming from the planners, makessure the foremen are up to date, generates the anticipated delay re-port for the planners, helps break bottlenecks, and so forth.5
The last point, breaking bottlenecks, brings up another cutover issue: overloads This can be a problem because, again, Ca-pacity Requirements Planning isn’t operational yet Overloads arebad because the work won’t get through on time Underloads are al-
post-5 This person can also make progress on plant floor cleanup activities during this period Example: getting some of the large queues off the floor and back into the stockroom The newly generated MRP priority information can help a lot in this job.
Trang 13most as bad because people will run out of work and get a negativefeeling about ERP.
This can be a major problem, and one that the project teamneeds to be aware of and follow very closely Planning ahead canhelp a lot Some companies have done a pre-cutover dry run or sim-ulation to see what’s likely to occur Contingency plans can help alot: plan A might be what to do in case of an overload; plan B for
an underload
Once again, that key plant person mentioned earlier can be a bighelp—by eyeballing the queues, talking to the foremen about theirproblems, talking to the material planners about what ERP shows iscoming soon, breaking the bottlenecks, and making certain the plantdoesn’t run out of work During this tricky transition period, dowhatever possible to anticipate problems Identifying them ahead oftime can minimize their impact
The buyers have a similar role to play with their suppliers Theyneed to follow up closely with their suppliers, learn which orders willnot be shipped on time, and communicate these to the planners viathe anticipated delay report
There’s also a potential capacity problem with suppliers SinceMRP is now involved in planning orders, the orders might not becoming out in the same pattern as before The company could inad-vertently be creating severe overloads or, just as bad perhaps, severeunderloads at key suppliers The buyers need to stay in close contactwith these suppliers to solve these kinds of problems should theyarise
The three most important things the people can do during this riod are:
com-234 ERP: M I H