GATE Mechanical: Industrial Engineering and Operations Research, previous-year questions solved
Every industrial engineering question from the last four GATE Mechanical papers that has an official key: EOQ, reorder point, quantity discounts and MRP, linear programming (shadow prices, the dual, a simplex pivot), the assignment and transportation problems, CPM and PERT, exponential smoothing and forecast errors, an M/M/1 queue and critical-ratio scheduling. Each one has a step-by-step solution that matches the official answer key, the trap that catches most people, and a note on where the method is used in plants, warehouses and project offices.
Questions from official GATE ME papers (IIT Kharagpur 2022, IISc Bengaluru 2024, IIT Roorkee 2025, IIT Guwahati 2026); solutions by Skill-Lync. · Updated 16 Nov 2026 · 20 questions
Which one is NOT a form of inventory: raw materials, work-in-process, finished goods or CNC milling machines?
- (A)Raw materials
- (B)Work-in-process materials
- (C)Finished goods
- (D)CNC Milling Machines
Answer: (D) Official key: D
- Inventory is stock held to be used up or sold: raw materials waiting for production, work-in-process (partly made parts between operations) and finished goods waiting to be shipped. Spares and consumables (MRO stock) also count.
- A CNC milling machine is not consumed or sold; it is used again and again to make the product. It is a fixed asset (capital equipment), not inventory: option (D).
The split matters in the books and on the shop floor: inventory sits in working capital and is counted in stock-takes and ERP stock reports, while machines are fixed assets that are depreciated. Lean programmes attack work-in-process first, because parts waiting between machines hide quality problems and tie up cash.
Demand 1000 parts a year, lead time 5 days, orders by EOQ: at what stock level should you reorder?
Answer: 14 parts Official key: 13 to 15
- The reorder point is the stock that covers demand during the lead time: .
- Daily demand: parts per day, so , which rounds to parts.
The EOQ ( parts) tells you how much to order, not when; the unit price and ordering cost are distractors here. A 360-day year gives 13.9, also inside the key; using 250 working days gives 20, outside it.
ERP systems (SAP's reorder point planning, for example) trigger a purchase requisition exactly this way when stock falls to the reorder level. In practice a safety stock is added on top, sized from how much demand and supplier lead time vary, so the line does not stop when a delivery is late.
Ordering 25% more than the EOQ: by what percentage does the total inventory cost rise?
- (A)2.5
- (B)5
- (C)0
- (D)12.5
Answer: (A) Official key: A
- Total cost . At the EOQ the two terms are equal, and .
- For :
- With : , a rise of 2.5%: option (A).
The demand, ordering cost and holding cost drop out completely. With per gear per year you get and rupees a year, but the ratio is 1.025 whatever they are; even using the monthly figure by mistake gives the same 2.5%. The flat cost curve near the EOQ is why ordering in round lots barely costs anything.
Because the cost curve is so flat near the optimum, buyers round the EOQ to pallet, coil or box quantities without worrying: being 25% off costs only 2.5% more. What does hurt is a large error, for example ordering half or double the EOQ (25% more cost).
3000 units a year with price breaks at 500 and 1000 units: what is the minimum total annual cost?
| Order quantity | Unit price (₹) |
|---|---|
| 1 to 499 | 9.0 |
| 500 to 999 | 8.5 |
| 1000 or more | 8.0 |
Answer: ₹26050.0 Official key: 26000.0 to 26100.0
- Holding cost depends on price (), so find the EOQ at each price: with , .
- ₹8.0: , below 1000, so not allowed at this price; the cheapest quantity that gets ₹8.0 is 1000.
- ₹8.5: , inside 500 to 999: feasible. ₹9.0: , just above the 499 limit, so it cannot beat ₹8.5.
- :
at :
at : ₹26050.0.
Stopping at the feasible EOQ (₹27249.3) is the trap: with an all-units discount the cost must also be checked at every higher price break.
Purchase teams run exactly this comparison when a supplier quotes price breaks, then add what the textbook leaves out: storage space, shelf life, cash tied up and the risk of an engineering change making the stock obsolete. A bigger lot that looks cheaper on paper is often rejected for those reasons.
Holding cost ₹0.50 per unit per month, ordering cost ₹550, demand 10000 a year: what is the EOQ?
Answer: 1354 units Official key: 1353 to 1355
- Put the holding cost on the same time basis as the demand: rupees per unit per year.
- units per order.
Using ₹0.50 directly with the annual demand gives 4690 units: the classic unit slip. Monthly demand (833.3) with the monthly holding cost gives the same 1354, because only the ratio matters.
Distributors and stockists recompute lot sizes when interest rates or warehouse rents change, because holding cost is mostly the cost of money tied up in stock plus storage. A common rule of thumb puts annual holding cost at roughly 20-30% of the item's value.
Component A appears at two levels of the product tree and 50 are in stock: how many more are needed for 10 units of P?
Answer: 110 units Official key: 110 to 110
- A is used in two places, so add both paths through the tree.
- Directly under P: .
- Under D: .
- Gross requirement ; net requirement units.
The common slip is adding the numbers in the boxes without multiplying down the tree, which gives ; quantities in a product structure are per unit of the parent, so they multiply along each branch.
This is a bill-of-materials explosion, the core of material requirements planning (MRP). ERP systems run it every night for thousands of part numbers: they explode the production plan through multi-level BOMs, net off stock and open orders, and offset each requirement by its lead time to create purchase and work orders.
In linear programming, what is the shadow price of a resource that is not fully used?
- (A)positive
- (B)negative
- (C)zero
- (D)infinity
Answer: (C) Official key: C
- The shadow price of a resource is how much the optimal objective improves if one more unit of that resource is available.
- If some of the resource is already left over (its slack is positive), one more unit changes nothing, so its shadow price is zero: option (C).
This is complementary slackness: for every constraint, . A resource can have a non-zero shadow price only if it is used up completely.
Planners read shadow prices straight off the solver report (Excel Solver's sensitivity report, or the dual values in any LP solver) to decide where extra capacity pays: overtime on the bottleneck machine has a positive shadow price, more hours on a machine that already sits idle are worth nothing.
Write the dual of a two-variable minimisation LP with four ≥ constraints
, , , and represent the corresponding dual variables.
Which one of the following options represents the correct form of ?
- (A)Maximize subject to , ,
- (B)Maximize subject to , ,
- (C)Maximize subject to , ,
- (D)Maximize subject to , ,
Answer: (A) Official key: A
- A minimisation with constraints and has a maximisation dual with constraints and . The right-hand sides (0.4, 0.6, 2.0, 1.8) become the dual objective coefficients: all four options agree on that.
- Each primal variable gives one dual constraint, built from its column of coefficients. Column of : 0.1, 0, 0.1, 0.2 with cost 0.07, so .
- Column of : 0, 0.1, 0.2, 0.1 with cost 0.05, so .
- That is option (A). (D) swaps the costs 0.07 and 0.05; (B) and (C) put in the constraint, i.e. they read the rows instead of the columns.
Dual values are the prices of the constraints, so a cost-minimising blend or diet problem like this one (minimum nutrient or property levels at least cost) tells the planner what each specification really costs. Feed mills, refineries and alloy melt shops use these numbers to argue about which spec limits are worth keeping.
One simplex pivot from a given dictionary: what is the objective at the next basic feasible solution?
- (A)
- (B)
- (C)
- (D)
Answer: (A) Official key: A
- At the non-basic variables are zero and . Both and have negative coefficients, so raising either one lowers .
- Raise (with ): limits it to 2, to 3.5, does not limit it. So , leaves, and .
- Raise instead: only limits it, , . Entering improves more (−4 against −3), and it also has the steepest coefficient, so both rules pick .
- Rewrite with non-basic: , so : option (A).
(B) is the dictionary after the weaker pivot (enter , leaves). (C) uses the wrong leaving variable, and (D) has the wrong constant. The term says the method is not finished: enters next.
Commercial LP solvers (Gurobi, CPLEX, HiGHS) still run simplex variants for most production planning, blending and scheduling models; the pivot logic is the same, just with sparse matrix tricks for models with millions of variables. Knowing what one pivot does is what lets you read a solver log or a degenerate, stalling model.
Four suppliers and four retailers, all 300 units: what is the minimum transportation cost?
| Supplier ↓ Retailer → | I | II | III | IV | Capacity |
|---|---|---|---|---|---|
| A | 11 | 16 | 19 | 13 | 300 |
| B | 5 | 10 | 7 | 8 | 300 |
| C | 12 | 14 | 17 | 11 | 300 |
| D | 8 | 15 | 11 | 9 | 300 |
| Demand | 300 | 300 | 300 | 300 |
Answer: ₹12300 Official key: 12300 to 12300
- Supply (1200) equals demand (1200), so the problem is balanced.
- Every supplier and every retailer has exactly 300 units, so an optimal plan sends each supplier's whole 300 to one retailer: the problem is an assignment problem with every cost multiplied by 300. (The corner points of this transportation problem are exactly the one-to-one plans.)
- Check all 24 one-to-one plans, or use the Hungarian method: the cheapest is unique, A → I (11), B → III (7), C → II (14), D → IV (9), total per unit; the next best costs 42.
- Minimum cost ₹12300.
The north-west corner start (A-I, B-II, C-III, D-IV) costs ₹14100: a feasible plan, not the optimum. The optimum can be confirmed with MODI (u-v) prices: no unused route has a negative reduced cost.
Distribution planners solve far larger versions of this (plants to warehouses to dealers, with capacities, lead times and truck sizes) in supply chain planning software. The classic transportation model is still the core, which is why GATE keeps asking it.
An assignment problem's optimum is 500 min; Job 4 now takes 20 min longer on every machine: what is the new optimum?
Answer: 520 minutes Official key: 520 to 520
- Every feasible assignment uses exactly one cell of Job 4's row.
- Adding 20 to that whole row therefore adds exactly 20 to the total of every assignment, so the ranking of assignments does not change and the same assignment stays optimal.
- New minimum minutes.
Adding 20 for each of the four machines () is the trap: only one machine ends up doing Job 4. This row-constant property is also why the Hungarian method may subtract the row minimum from every row without changing the answer. (We also checked it numerically on 200 random 4 × 4 problems.)
Shop-floor dispatching, crew rostering and allocating parts to suppliers are all assignment problems. Planners rely on the same property when they adjust a whole row or column, for example a slower operator or an older machine, and know the optimal pairing does not move.
Eleven activities A to K: what is the latest finish time of activity G?
| Activity | Time (hours) | Immediate predecessors |
|---|---|---|
| A | 2 | ─ |
| B | 3 | ─ |
| C | 2 | ─ |
| D | 4 | A |
| E | 5 | B |
| F | 4 | B |
| G | 3 | C |
| H | 10 | D, E |
| I | 5 | F |
| J | 8 | G |
| K | 3 | H, I, J |
- (A)5
- (B)10
- (C)8
- (D)9
Answer: (B) Official key: B
- Forward pass (earliest finish): A 2, B 3, C 2, D 6, E 8, F 7, G 5, H , I 12, J 13, K . The project takes 21 hours; the critical path is B-E-H-K.
- Backward pass: K must finish by 21, so it must start by 18. J's only successor is K, so J must finish by 18 and start by .
- G's only successor is J, so the latest finish of G is 10 hours: option (B).
5 is G's earliest finish (forward pass only). G has hours of float, so a delay of up to 5 hours on C, G or J together does not delay the project.
Every project schedule in MS Project or Primavera P6 is computed with these forward and backward passes. Total float is what planners watch: when it hits zero on a path, that path has become critical and any slip there moves the delivery date.
Five activities with optimistic, most likely and pessimistic times: what is the expected project time?
Time estimates (in weeks): optimistic time , most likely time , pessimistic time .
| Activity | Immediate predecessor(s) | |||
|---|---|---|---|---|
| A | 4 | 5 | 6 | None |
| B | 1 | 3 | 5 | A |
| C | 1 | 2 | 3 | A |
| D | 2 | 4 | 6 | C |
| E | 3 | 4 | 5 | B, D |
Answer: 15 weeks Official key: 15 to 15
- PERT expected time of each activity:
- , , , , . (Each estimate here is symmetric, so equals the most likely time.)
- Paths: A-B-E ; A-C-D-E .
- The expected completion time is the longest path: weeks (critical path A-C-D-E).
Three-point estimates are still how project managers price uncertainty in R&D, new-product launches and plant shutdowns. Modern tools go one step further and run Monte Carlo simulation on the network, because PERT's single critical path tends to understate the risk when several paths are nearly critical.
Eight activities A to H: what is the slack of activity D?
| Activity | Immediate predecessor | Duration (hours) |
|---|---|---|
| A | - | 4 |
| B | A | 8 |
| C | A | 5 |
| D | B | 2 |
| E | B | 7 |
| F | C | 6 |
| G | D | 3 |
| H | E, F, G | 9 |
Answer: 2 hours Official key: 2 to 2
- Forward pass (earliest start, finish): A 0-4, B 4-12, C 4-9, D 12-14, E 12-19, F 9-15, G 14-17, H starts at and ends at 28.
- Backward pass: H must start by 19, so G must finish by 19 and start by 16; D's only successor is G, so D must finish by 16 and start by 14.
- Slack of D: hours.
The critical path is A-B-E-H (28 hours). The path through D (A-B-D-G-H, 26 hours) is 2 hours shorter, which is the same 2 hours of slack shared by D and G.
Slack tells a site or plant manager which jobs can be shifted to level out labour and crane demand without moving the finish date. Resource levelling in scheduling software uses exactly this float, starting with the activities that have the most of it.
A PERT network of eight activities from node 1 to node 6: what is the expected project length?
| Activity | |||
|---|---|---|---|
| 1−2 | 2 | 4 | 12 |
| 1−3 | 2 | 4 | 6 |
| 1−4 | 3 | 4 | 11 |
| 2−5 | 3 | 6 | 9 |
| 3−4 | 2 | 5 | 14 |
| 3−5 | 3 | 3 | 3 |
| 4−5 | 3 | 6 | 15 |
| 5−6 | 2 | 5 | 8 |
Answer: 22 weeks Official key: 22 to 22
- for each activity: , , , , , , , .
- All paths from node 1 to node 6, in weeks: : 22; : 17; : 16; : 12.
- The expected project length is the longest path: weeks, along 1-3-4-5-6.
Using the most likely times alone gives 20 weeks, which is the trap: activities 3-4 and 4-5 have long pessimistic tails (14 and 15 weeks), and PERT's weighted mean pulls them up.
Long pessimistic tails are typical of tasks that depend on outside parties, such as tooling from a vendor, certification tests or customs clearance. Planners flag activities with a wide spread between and for early follow-up, because they drive both the expected date and its uncertainty.
January sales were 120 and the forecast was 20 too low: what is February's forecast with α = 0.2?
Answer: 104 units Official key: 104 to 104
- Underestimated by 20 means the January forecast was .
- units.
Reading it as a forecast of 140 gives 136: an underestimate means the forecast was below the actual.
Exponential smoothing (and its trend and seasonal versions, Holt and Holt-Winters) is the default forecasting method in most demand planning software, because it needs only the last forecast and the latest actual. A small α like 0.2 gives a steady forecast that ignores one-off spikes.
Ten periods of demand and forecast: what are the mean forecast error and the mean absolute deviation?
| Period | Actual demand | Forecast |
|---|---|---|
| 1 | 425 | 427 |
| 2 | 415 | 422 |
| 3 | 420 | 416 |
| 4 | 430 | 422 |
| 5 | 427 | 423 |
| 6 | 418 | 420 |
| 7 | 422 | 419 |
| 8 | 416 | 418 |
| 9 | 426 | 430 |
| 10 | 421 | 415 |
- (A)0.8 and 42.0
- (B)0.8 and 4.2
- (C)8.0 and 42.0
- (D)8.0 and 4.2
Answer: (B) Official key: B
- Errors (actual − forecast), period 1 to 10: , , , , , , , , , .
- Mean forecast error .
- Mean absolute deviation .
- 0.8 and 4.2: option (B).
The options with 8.0 and 42.0 are the sums, not the means. A small mean error with a much larger MAD means the forecast is nearly unbiased but still misses by about 4 units a period in either direction.
Demand planners track both numbers every month: the mean error (bias) shows whether the forecast is systematically high or low, and MAD sizes the safety stock (for normally distributed errors the standard deviation is about 1.25 × MAD). A tracking signal, the running sum of errors divided by MAD, flags a forecast that has drifted.
Actual demand 500 then 635, January forecast 250, α = 0.7: what is the March forecast?
Answer: 572 units Official key: 572 to 572
- , applied twice.
- February: .
- March: units.
Skipping the February update and smoothing 635 against the January forecast gives 519.5. A high α (0.7) makes the forecast chase the latest actual: it closed most of a 250-unit gap in one month.
Foundries and forge shops plan melt schedules and raw material from forecasts like this. A high smoothing constant reacts quickly to a genuine jump in orders but also to noise, so planners usually keep α between 0.1 and 0.3 and switch to a trend model when demand is clearly growing.
Arrivals 5 jobs per hour, mean service 6 minutes, single server: what is the probability the server is idle?
- (A)0.20
- (B)0.17
- (C)0.50
- (D)0.83
Answer: (C) Official key: C
- This is an M/M/1 queue. Arrival rate jobs/hour; a mean service time of 6 minutes means a service rate jobs/hour.
- Utilisation : the server is busy half the time.
- Probability the server is idle: : option (C).
Reading the 6 minutes as 6 jobs/hour gives (option D) and (option B).
The same formulas size inspection stations, tool cribs, maintenance crews and AGV fleets. The key lesson is how fast waiting grows with utilisation: in an M/M/1 queue the average number in the system is , 1 job at 50% utilisation but 9 at 90%, which is why bottleneck machines are never planned at 100%.
On day 20, which jobs have a critical ratio below 1?
| Job | Due date | Work remaining (in days) |
|---|---|---|
| M | 26 | 5 |
| N | 24 | 10 |
| O | 28 | 10 |
| P | 31 | 5 |
- (A)Job N and Job O
- (B)Job M and Job N
- (C)Job P
- (D)Job O and Job P
Answer: (A) Official key: A
- Critical ratio is the time left until the due date divided by the work remaining:
- M: ; N: ; O: ; P: .
- Below 1 (behind schedule): N and O, option (A).
A CR below 1 means the job cannot finish on time at the normal pace; the CR rule runs the lowest ratio first, so the order would be N, O, M, P.
Job shops and tool rooms use critical ratio in their dispatch lists because it updates itself every day: as the due date approaches, a job's ratio falls and it climbs the list. Manufacturing execution systems show it next to each work order so supervisors can see which jobs need overtime.