Specimen paper

A pharmaceutical packer in Amman runs blistering, cartoning and serialisation in sequence. First-pass yields are 90 per cent, 95 per cent and 80 per cent respectively. What share of units passes all three steps with no rework at any of them?

• A

  68.4 per cent, the three first-pass yields multiplied together.

  Correct: rolled throughput yield is the product of the step yields, and 0.90 by 0.95 by 0.80 gives 68.4 per cent.

• B

  88.3 per cent, the three first-pass yields averaged across steps.

  This is the mean of the three yields, and it comes from treating the steps as alternatives that a unit might pass through rather than a sequence every unit must survive. An average would answer a different question, namely how the three steps compare with one another. It would be the right figure only if each unit went through one step and not all three.

• C

  80.0 per cent, the lowest first-pass yield of the three steps.

  This borrows the bottleneck rule from capacity, where the slowest step genuinely does set the rate, and applies it to yield, where it does not. It assumes blistering and cartoning reject nothing, when between them they lose about 14 per cent of what enters. It would be right only if the other two steps ran at 100 per cent first-pass yield.

• D

  65.0 per cent, the three yield losses deducted from a hundred.

  This adds the losses of 10, 5 and 20 points and subtracts them, which treats compounding as addition. It always understates the answer, because a unit already scrapped at blistering cannot be scrapped a second time at cartoning; here it is 3.4 points low, and the gap widens with every extra step. It would be a fair approximation only where the losses are very small.

At a Riyadh utility, requisitions under SAR 5,000 need three signatures and take four days to clear. Two of the three approvers have never rejected one. The IT team proposes a workflow tool that routes the same three approvals electronically and would cut the delay to one day. What is the strongest objection?

• A

  Two of the three approvals decide nothing and should be removed first

  Correct: two of the approvals make no decision, so they should be removed before anything is spent on routing them faster.

• B

  Approvers will still work in batches, so one day is an optimistic figure

  This is a fair criticism of the estimate, since people work their queues in batches whether the queue is paper or electronic, and it is the answer a candidate reaches by scrutinising the benefit rather than the design. It accepts all three approvals as necessary and argues only about speed. It would be the strongest objection if all three approvers genuinely decided something.

• C

  Electronic routing weakens the audit trail the signed form provides

  This assumes paper is the stronger record, when a workflow system captures identity, time and sequence better than a signature does. The instinct behind it is sound governance applied to the wrong medium. It would carry weight only where a regulator specifically requires a wet signature on the document.

• D

  The tool will not pass data to the finance ledger as it is installed

  Integration is a genuine implementation risk and belongs on the project's risk log, which is why it reads like a serious objection. It says nothing about whether the three approvals should exist, so it cannot be the strongest objection to the proposal itself. It would be decisive only if the business case rested on the ledger feed.

A Lahore pharmaceutical plant takes 70 minutes to change its tablet line to another product. Of that, 12 minutes is fetching tooling from stores, 9 minutes is finding and loading the recipe and 6 minutes is collecting labels; the rest is stripping, cleaning and first-off checks. Which action gives the largest gain for the least cost?

• A

  Add two more operators to the changeover so tasks are done at once

  This puts labour on to work that need not happen while the line is down at all, so it pays every changeover for a gain that staging gives away free. Parallelising is a legitimate later step, applied to internal work that is genuinely internal. It would be the right answer if the fetching had already been moved off-line and the remaining stripping and cleaning could be split between people.

• B

  Stage tooling, recipe and labels at the line while it is still running

  Correct: this converts 27 of the 70 minutes from internal work to external work, at the cost of a trolley and a checklist.

• C

  Buy a second set of tooling so cleaning is done away from the line

  This is orthodox changeover practice and often right, which is what makes it tempting, but it spends capital on a problem that staging solves for nothing, and the question asks for the largest gain at the least cost. It also leaves the fetching and searching untouched. Consider it once the external work has gone and off-line cleaning is the largest internal element remaining.

• D

  Time each element and set a target time for the whole changeover

  Measurement is a sound preparatory step, and the team has plainly done enough of it already to produce the breakdown in the stem. A target changes no element of the work, and its usual effect is to make crews hurry internal steps rather than remove them. It would be the right answer only if nobody yet knew where the 70 minutes went.

A visa-processing centre in Doha opens five assessment counters for 480 minutes a day, of which 60 minutes are scheduled breaks when no applications are handled. The centre must clear 420 applications a day, and an assessor currently spends an average of 75 seconds on an application. What takt time should the counter work be designed around?

• A

  Sixty seconds, being available counter minutes divided by applications.

  Correct: 480 opening minutes less 60 unstaffed break minutes is 420 available minutes, and 420 minutes against 420 applications is 60 seconds.

• B

  Seventy-five seconds, being the average time an assessor spends on a file.

  This offers cycle time as takt, taking the rate the process happens to produce instead of the rate demand requires. The two are equal only by coincidence, and here the gap of 15 seconds is precisely the finding the calculation exists to reveal. It would be the right answer to a question about how long an assessor currently takes, which is the comparison you make after takt is known.

• C

  Sixty-nine seconds, being the full opening time divided by applications.

  This counts the hour of breaks as available time, which builds the design on minutes when nobody is at a counter and sets a pace the centre can never hold across a full day. Available means staffed, not open. It would be right if the counters ran through breaks on a relief rota.

• D

  Twelve seconds, being the takt shared across the five counters on duty.

  This divides takt among the servers, confusing the pace of the process with the pace of one station. Takt is a single demand rate and does not change with staffing; staffing is what you decide afterwards, by comparing the 75 seconds of work content with the 60-second takt. Twelve seconds would be the correct takt only if the centre had to clear 2,100 applications a day.

A Jakarta distributor sells household goods made by a plant that supplies it weekly. Consumer sales at the shelf vary by about 5 per cent week to week; store orders vary by 15 per cent; depot orders on the plant vary by 30 per cent; the plant's own production schedule varies by 45 per cent. The plant's demand planner asks the retailer to improve its forecast. What does the pattern indicate?

• A

  Consumer demand is far more volatile than the shelf figures suggest

  This reads the amplification as evidence that the shelf figure conceals the real swings, which is the relationship exactly backwards: the shelf is the least distorted signal anyone in the chain possesses. Somebody choosing it has assumed variation must originate with the customer. It would be right only if the 5 per cent were an aggregation artefact hiding volatility underneath, and nothing in the stem points that way.

• B

  The plant's safety stock is too small for the demand it actually faces

  This treats the amplified orders as a given and buys cover against them, which is expensive to hold, quick to obsolete and grows with the very batching that caused the problem. It is a defensible interim measure while the ordering rules are changed, but it answers the symptom. It would be the right first move only if the variation were genuinely external and could not be reduced.

• C

  Variability is added at each handover, largely by batching and rounding

  Correct: the variation grows at every handover away from the consumer, which is the signature of order batching, case and pallet rounding and weekly review cycles.

• D

  The retailer's forecasting is the source and it should be corrected first

  This is the planner's assumption in the stem restated. The retailer does add variation, from 5 per cent at the shelf to 15 per cent in its orders, but that comes from its ordering rules rather than its forecast, and two further amplifications happen after the orders leave it. Correcting the forecast would spend commercial goodwill and leave the batching in place; it would be right only if the retailer's orders were the last distortion in the chain.

At a Penang circuit-board assembler, reflow is the constraint and pulls 60 boards a minute. The buffer conveyor ahead of it holds 1,800 boards. The feeding step stopped for 50 minutes on Tuesday and for 20 minutes on Thursday; it runs 25 per cent faster than reflow pulls, so the buffer was full again before each event. Contribution is USD 3 a board. What did the two stoppages cost?

• A

  USD 12,600, from the full 70 minutes the feeding step was stopped

  This prices every upstream minute as though it were a constraint minute, which is the error a downtime report actively invites: it lists both events in identical units and says nothing about what stood between them and reflow. It would be the right figure only if there were no buffer at all, or if the feeding step were itself the constraint.

• B

  USD 3,600, from the 20 minutes that outran the buffer on Tuesday

  Correct: 30 minutes of buffer cover absorbs Thursday entirely and 30 of Tuesday's 50 minutes, leaving 20 minutes, 1,200 boards and USD 3,600.

• C

  USD 7,200, from the 70 minutes less one buffer of 30 minutes

  This adds the two stoppages together and then deducts a single buffer, which would be sound if the events were consecutive with no chance to recover between them. The stem states that the buffer was full again before each event, so each stoppage meets 30 fresh minutes of cover and has to be tested separately. Aggregating first is the commonest way of getting this calculation wrong.

• D

  Nothing, since the faster feeding step refilled the buffer each time

  This confuses what protective capacity does between events with what it can do during one. The 25 per cent margin is exactly why the buffer was full again by Thursday, but it cannot help while the feeding step is stopped and the buffer is draining. It would be right if both stoppages had been shorter than the 30 minutes of cover, as Thursday's was.

At a hospital pharmacy in Doha, two look-alike vials of different strengths sit in adjacent bins and have been picked in error four times this year. Four proposals are on the table. Which one deals with the error most reliably?

• A

  Fit bins shaped so that each accepts only one of the two vials

  Correct: shaping the bins makes the wrong pick physically impossible, so the error is removed at the point it would be made rather than caught afterwards.

• B

  Fix a red warning label on both bins naming the strength clearly

  A label instructs, which is the weakest control available, because it works only when the person reads it at the moment of the pick. The people most likely to miss it are exactly the ones under pressure at the end of a shift, which is when the error occurs. It would be the best available answer only if no physical or detection option existed at all.

• C

  Add a second pharmacist to check each pick before it is issued

  This is the closest near miss: it looks like prevention but is detection, permitting the error and then relying on a busy person to notice it, at the cost of an extra check on every pick for ever. Double-checking also carries a false-reject rate and a well-documented decay in attention. It would be reasonable as a temporary measure while the bins are being made.

• D

  Check the vial against the prescription on the ward before use

  This also detects rather than prevents, and it does so further downstream, after the vial has left the pharmacy and closer to the patient. Moving a control away from the point of error is the wrong direction of travel, however conscientious the checker. It should stay in place as a final safety net whatever else is done, but it cannot be the plan.

A filling line at Indus Dairy in Karachi charts fill volume shift by shift. The two lines drawn on the chart are the customer's tolerance of 995 ml and 1,005 ml. The process has a natural spread of 998 ml to 1,002 ml and no point has ever crossed a line. The quality manager concludes that the chart proves the filler is behaving. What is the flaw?

• A

  The chart shows tolerance breaches, not changes in the process

  Correct: with tolerance lines in place of calculated limits, the chart can only signal a breach of specification and is blind to any change in the filler itself.

• B

  The chart should carry limits tightened well inside the tolerance

  Tightening the lines feels prudent and still uses the wrong quantity, because a line placed by judgement signals when somebody decided it should rather than when the process changed. Limits have to be calculated from the filler's own variation; there is no correct place to put an arbitrary line. It would be right only if the tolerance were being used as a genuine action limit for a different purpose, which is not a control chart.

• C

  The chart should be redrawn from the overall spread of all shifts

  This is the right instinct applied to the wrong data. Control limits come from short-term variation, such as the movement between consecutive shifts; a spread computed across all shifts already contains any drift or step change, so the limits are inflated by the very signal they exist to detect and the chart falls silent. It would be sound only where the whole run genuinely is one rational subgroup.

• D

  The chart needs a far longer run of points before its lines mean much

  Limits do become more stable with more subgroups, so this sounds like careful practice, and it is beside the point here. The lines are the wrong lines, and a thousand further readings against a tolerance would still tell you nothing about the filler. It would be the answer if calculated limits had been drawn from too few points.

A Gujarat pipe extrusion plant is loaded for 100 hours a week and runs for 80, giving availability of 80 per cent. The engineering manager reclassifies the five hours a week of recurring die-heater faults as planned maintenance, so loading time becomes 95 hours and availability is reported at 84.2 per cent. Tonnes despatched are unchanged. What does the improvement represent?

• A

  Nothing: the plant sets the denominator, and no more pipe was made.

  Correct: the entire reported gain comes from shrinking the denominator, and unchanged tonnage confirms that nothing physical has changed.

• B

  A real gain: planned maintenance is properly outside loading time.

  The general rule is correct, which is exactly what makes this the tempting answer, but the exclusion is for work scheduled in advance, not for breakdowns relabelled afterwards because they happen often enough to be predictable. The hours were unplanned when they were lost. It would be a real gain if the die-heater work had been planned, slotted into a window and carried out to that schedule.

• C

  A partial gain: the stops are planned now, so they can move off-peak.

  This awards credit for something that has not happened yet. Moving stoppages into an idle window would be a genuine improvement, but the stops occurred exactly when they occurred and only the label changed. It would be right if the plant had actually rescheduled the work, and then the evidence would be a change in tonnes.

• D

  A measurement error: those hours belong in performance loss instead.

  This spots that the accounting is wrong and then puts the hours in the wrong bucket. Stoppage time is an availability loss; performance loss is running below rated speed while the machine is up. Filing five hours of downtime as speed loss would leave availability overstated in precisely the same way.

A Busan home-appliance maker ships 1,000 orders a year to four retail groups at an average order value of USD 50,000, so annual order value is USD 50m. The contract charges 2 per cent of order value on any order that is short or late. Units delivered on time and in full ran at 97 per cent last year. What can the finance director conclude about penalty exposure from that 97 per cent alone?

• A

  Anywhere from USD 30,000 to USD 1m, depending on how it is distributed.

  Correct: the same 97 per cent on units is consistent with 30 badly short orders or with all 1,000 slightly short, so only a range can be stated.

• B

  About USD 30,000, being 2 per cent of value on the 3 per cent of orders.

  This converts a percentage of units straight into a percentage of orders, which holds only if shortfalls arrive whole orders at a time. It is the natural reading, and it happens to select the cheapest of the possible worlds, which makes it a comfortable and dangerous number to take to a board. It would be correct if the report showed 30 orders breached.

• C

  About USD 1m, being 2 per cent of the annual order value across the four.

  This applies the penalty rate to the whole order book, which assumes every one of the 1,000 orders was short or late. That is the far end of the range rather than an estimate, and it overstates the likely charge as confidently as the previous option understates it. It would be right only in the worst of the distributions consistent with 97 per cent.

• D

  About USD 1.5m, being the value of the units that were not shipped on time.

  This is 3 per cent of annual order value, which is the sales value of what did not ship, not a contractual charge. It confuses revenue forgone or delayed with the penalty, and the two are separate losses that could both fall in the same year. The contract charges 2 per cent of order value on breached orders and turns on nothing else.

A Dubai freight terminal builds a skills matrix for the night gate. Dangerous goods checker: 1 required, 1 competent. Reach truck driver: 2 required, 6 competent. Documentation clerk: 2 required, 5 competent. Which reading of this matrix is correct?

• A

  The dangerous goods checker is a single point of failure on nights

  Correct: one competent person against one required is a cover ratio of 1.0, so a single absence or resignation stops that activity.

• B

  The reach truck drivers are the constraint, being the busiest role

  This picks the role with the most people in it and calls that a constraint. Headcount says nothing about cover, and six against two required is the strongest line on the matrix rather than the weakest. A constraint is set by capacity against demand, which is a different analysis and not one this matrix supports.

• C

  The gate is well covered, since competent people exceed those required

  This totals twelve competent against five required and pronounces the gate safe, which is the specific aggregation error a skills matrix exists to prevent. Competence is not transferable between roles, so a surplus of reach truck drivers does nothing about a missing dangerous goods checker. It would be right only if any competent person could cover any post.

• D

  The documentation clerks carry the greatest risk, at 2.5 per post

  This reads the ratio backwards, treating the second-best cover on the gate as the largest exposure, probably by reading the largest available number as the worst news. For 2.5 per post to be a risk you would need a reason why more cover is worse, and none exists. It would be right only if the figure were 0.5 rather than 2.5.

A team leader at a Philippine business-process centre has known for four months that one agent transfers difficult calls to protect his handling time. At the annual review, the leader raises it between two compliments. The agent leaves believing the review went well. What is the primary defect?

• A

  The point was made without a written record for the agent

  A record matters, particularly if the behaviour continues and the matter later becomes formal, so this is a genuine omission. It is secondary, because a written note of a message the agent did not hear as criticism only documents the ambiguity. It would be the primary defect if the conversation itself had been timely and unambiguous.

• B

  The point was made without the number of calls transferred

  Taking the data would have made the point concrete and hard to dispute, and you should always bring it. The agent did not miss the message for want of evidence, though; he missed it because it arrived four months late and softened. Evidence sharpens a message that has been delivered clearly, and cannot repair one that has not.

• C

  The point was made without a date set to review the change

  A review date is what turns feedback into a commitment, and without one nothing will be checked, so this too is a real omission. It is downstream of the main fault: there is little point diarising a review of a change the agent does not know he has been asked to make. It would be the primary defect if he had left the room clear about what must change.

• D

  The point was saved for the review and wrapped in praise

  Correct: holding the point for four months and then cushioning it between compliments is what allowed the agent to leave believing the review had gone well.