One Robot for Two Machines vs. Separate Loading

The fastest payback does not come from the cheapest setup. It comes from the option that produces more saleable good parts for every tenge invested. One shared robot often wins with long, similar machining cycles, a stable part range, and moderate demand. Separate loading overtakes it when machines start waiting for the robot, changeovers stop the whole cell, or orders already exist for the extra output.

A supplier quote will not show this. You need a model that connects capital costs with a cycle timeline, equipment availability, gripper changes, zone guarding, and the real production calendar. If even one of these elements is replaced with an optimistic percentage, the payback period will look better than the line itself.

Output constrains payback more than robot price

Compare options by the annual cash flow from good parts, not by the cost of manipulators. The price difference is visible immediately, while lost output only appears after launch, when a machine waits several times a shift for a busy robot.

First determine what the business can sell in addition after automation. A part made above confirmed demand and left in the warehouse creates no contribution margin. It ties up metal, tooling, and working capital. Production capacity should therefore be limited by the lower of two figures: available cell output and order volume.

Calculate released labor separately. If, after launch, an operator can tend more machines, perform inspection, or prepare tooling, the business gains from reallocating working time. If that person still has to stand at the same cell for the whole shift because feeding is unreliable, measurements are manual, and chips need clearing, you cannot count their full salary as savings.

For assessing an existing line, the OEE method is useful. Vorne breaks it down into availability, performance, and quality. However, OEE covers only planned production time. An investment model must also add calendar utilization: the number of shifts with real orders. A cell with strong OEE and orders for one shift will not pay back equipment as quickly as a technically less perfect line running three shifts.

Check four separate indicators:

• technical capacity at the specified cycle times;
• good-part output after downtime and scrap;
• saleable output, limited by demand;
• net cash benefit after maintenance and consumables.

Combining these indicators into one percentage hides the cause of losses. A capital-investment decision needs their connection, not an attractive average.

First, distinguish options by feeding method

“Separate loading” must refer to specific equipment, because two robots, two gantry loaders, and two bar feeders have different economics. Comparing “one robot versus two systems” without describing the blank is almost useless.

A shared robot between machines takes a part from a buffer, serves both chucks, and unloads finished parts. It saves on the manipulator, controller, and some peripheral equipment, but it ties two machines to a shared command sequence and safety circuit.

A separate arrangement assigns a feeding device to each machine. These may be two robots or simpler mechanisms. It requires more equipment, but each machine gets an independent loading resource and can keep running while its neighbor is being changed over.

A third option is often overlooked: a shared robot for complex individual blanks and a simple dedicated feeder for a stable part on the second machine. This combination may cost less than two universal robots and cause fewer mutual stoppages than a single cell.

The popular advice to “buy a universal robot and find other tasks for it later” sounds reasonable because versatility is easy to show in a presentation. On the shop floor, moving it requires new tooling, a program, risk assessment, machine communication, and renewed acceptance testing. If the company has not budgeted this work and a window for the move, future versatility has zero value in the financial model.

For a bar part that is fully machined in one setup, a bar feeder is often the more direct solution. A robot makes sense when separate blanks must be oriented, setups changed, parts moved between operations, measurements taken, or different machine types served. First choose the physically suitable feeding method, then compare budgets.

Cycle balance sets the limit of a shared cell

One manipulator suits two machines when it completes every required operation before each machine's next request and retains time for variation. Matching nameplate cycle times guarantees nothing by itself.

Let P1 and P2 be the intervals between machine requests, and S1 and S2 the time the robot is busy serving them. Service includes approach, opening the door, removing the finished part, cleaning locating surfaces, loading the blank, clamping, closing the door, starting the cycle, and moving away. An initial check looks like this:

Загрузка робота = S1 / P1 + S2 / P2

Если результат >= 1, схема физически не успевает.
Если результат < 1, нужна проверка конфликтов запросов и запаса времени.

A value below one is necessary but not sufficient. Two machines can request service at the same time. The robot will choose one, and the other will wait even though average manipulator utilization looks acceptable. Variable inspection time, inconsistent chip removal, and sensor-based part searches make these conflicts worse.

Create a timeline for every part family. Record not only the automatic CNC cycle but also the operations around it:

• time from the machine signal to the start of robot movement;
• door opening and chuck operation;
• part exchange, air blow-off, and gripper check;
• travel between the buffer and each machine;
• measurement, marking, and stacking of the finished part.

Then run the sequence of events for at least several dozen cycles. Shift the start of the second machine, add typical time variation, and include one stoppage to replenish the buffer. A spreadsheet will already reveal queues. Complex paths require simulation from the integrator, but it should still begin with measured times.

Robot idle time is not always a loss. If both machines are cutting metal, the manipulator waiting is normal. A loss occurs when a machine finishes machining and the robot is busy, changing a gripper, or stopped by a shared safety signal. The model needs machine downtime caused specifically by feeding.

Calculate cash flow from good parts

A complete model must calculate investment, available output, and annual net benefit separately. Without these three components, payback becomes the robot price divided by an operator's salary, and that calculation rarely survives the first year of operation.

Capital costs should include the manipulator, pedestal or base, grippers, buffers, machine interfaces, automatic doors, chip removal, measurement, guarding, safety devices, programming, installation, training, commissioning, and an initial tooling reserve. For an import project, separately record the contract currency, delivery, customs costs, and payment terms. Do not substitute the equipment price for the cost of a working cell.

The annual calculation can use the following lines:

Доступные часы = часы по календарю × коэффициент загрузки заказами

Мощность годных деталей =
доступные часы × доступность × производительность
× идеальная скорость выпуска × доля годных деталей

Продаваемые детали = MIN(мощность годных деталей; подтвержденный спрос)

Годовой денежный эффект =
дополнительные продаваемые детали × маржинальный доход на деталь
+ реально высвобожденные затраты труда
- обслуживание
- энергия и сжатый воздух
- захваты и прочие расходные материалы
- ожидаемая стоимость аварийных простоев
- дополнительные расходы на контроль и логистику

Простой срок окупаемости = капитальные затраты / годовой денежный эффект

Contribution margin per part is revenue less the variable costs incurred to make that part. You cannot use total revenue. Depreciation should not be subtracted as a cash expense in a simple cash-flow calculation either, although finance will need it for the tax model.

For projects with a meaningful service life, add discounted cash flow, taxes, a payment schedule, and residual value. Compare net present value over the same horizon. Simple payback is useful as a filter, but it ignores money received after the investment has been recovered.

When choosing between shared and separate arrangements, one more calculation is needed:

Добавочные инвестиции = CAPEX раздельной схемы - CAPEX общей схемы
Добавочный поток = годовой поток раздельной схемы - годовой поток общей схемы
Добавочная окупаемость = добавочные инвестиции / добавочный поток

If the incremental cash flow is negative or zero, the more expensive option does not recover the difference at the stated level of demand. That is an honest answer even if its technical capacity is higher.

Guarding changes both the estimate and availability

The safety system must be designed for the entire robotic cell, so one robot for two machines does not mean half the safety cost. A larger zone may require longer guarding, several controlled entrances, safe interfaces for both machines, and a process for independent access to each workstation.

ISO 10218-2:2025 addresses industrial robot safety at the application and cell level, including integration, commissioning, operation, maintenance, and decommissioning. This draws a useful line of responsibility: the manipulator's certificate does not prove that the assembled line is safe. ISO 12100 requires hazards to be identified and risks reduced throughout a machine's life cycle. Assessment must therefore include setup, chip cleaning, clearing a jammed part, and troubleshooting, not just automatic operation.

A shared robot often uses one perimeter. It may cost less than two fully independent zones, but opening one door can stop both machines and the manipulator. Splitting zones with controlled access lets one buffer be serviced while the other section keeps working, but it complicates the safety logic and may require more floor space.

Collaborative operation does not eliminate risk assessment. A gripped metal part, sharp edges, a chuck, chips, and an automatic door create hazards regardless of the robot's permitted hand-contact force. Sometimes speed and force limits make the cycle so long that lost output removes any saving on guarding.

Ask the integrator not for a line item called “safety package,” but for a zone layout and a list of operating modes:

• who enters each zone and why;
• what equipment stops when each entrance is opened;
• how reset and restart are performed;
• where the operator changes containers and removes chips;
• how energy is isolated for repairs.

Calculate guarding cost together with the value of lost time during normal access. A cheap zone that stops two machines every time it is replenished costs a great deal over its service life.

Gripper changes can erase the saving

A shared robot loses out when tooling changes are frequent and each changeover stops both machines. Separate loading lets each workstation keep its gripper and program, so a higher initial cost can sometimes be recovered through better availability.

For similar parts, adjustable jaws or a dual gripper may be enough. A dual gripper removes a finished part and loads a blank in one visit, but it increases mass and reach. This affects allowable payload along the path, speed, collision risk, and access to the chuck. Checking only the total mass of the part and gripper is too crude: load must be calculated with the center of gravity included.

An automatic changer is useful when part geometries are incompatible. It adds a tool magazine, air and signal connections, locking checks, additional program positions, and new failure causes. A change that takes seconds will not help if the operator then spends twenty minutes moving containers, chuck jaws, and measuring fixtures.

Calculate the cost of changeovers this way:

Годовая потеря от переналадок =
число переналадок × длительность остановки ячейки
× денежный эффект рабочего часа
+ стоимость пробных деталей и повторного контроля

For a shared robot, the cash benefit per operating hour includes both stopped machines. With separate loading, the second machine can keep producing if the safety zones and material flow are genuinely independent.

Collect the actual batch distribution over several production periods. The average batch hides an unpleasant situation in which one high-volume part runs for a week while the remaining jobs cause several short changeovers a day. The model must calculate each family separately and then add the annual results.

A shared failure doubles the impact

One robot links the availability of two machines, so its failure creates a common stoppage. Two independent devices contain more components, but failure of one usually does not deprive the second machine of loading.

Do not limit the analysis to a failed robot arm. Shared points of failure can include the controller, safety programmable controller, zone scanner, air supply, buffer, machine vision system, and coordination program. A jammed gripper can cost far less than a robot, but for output, the duration of the stoppage matters more than the price of the failed part.

Break downtime down by consequence:

Expected downtime cost can be estimated from event frequency, average recovery time, and lost cash benefit per hour. Take these values from logs for similar equipment and the service contract terms. Catalog reliability says little about financial risk without technician arrival time, spare-parts availability, and a program recovery process.

There is a reverse side. Two systems mean two preventive-maintenance sets, more spare items, and more programs. If the business cannot maintain them in-house, operating costs may consume the gain from independence. The model should therefore separate failure frequency from the scale of consequences instead of assigning an arbitrary “reliability factor” to one option.

Three demand scenarios give different answers

The same equipment set can pay back faster or slower solely because of the number of saleable parts. Consider an illustrative calculation that can be replaced with data from your own shop.

The first machine has a 210-second cycle and the second a 260-second cycle. The shared robot is busy for 50 seconds when serving the first and 58 seconds when serving the second. Its initial calculated utilization is 50 / 210 + 58 / 260, or about 0.46. The margin appears sufficient, but the timeline must still confirm that there are no long conflicts.

The total cost of the shared cell, including guarding, grippers, and integration, is assumed to be 38 million tenge. Two separate systems cost 48 million tenge. These are not market quotations, but illustrative inputs. The production plan is 4,000 hours per year. After accounting for downtime, availability is assumed to be 0.86 and good-part rate 0.985 for the shared arrangement, and 0.90 and 0.99 respectively for the separate arrangement. The cycles already include normal speed reduction.

The calculation gives about 105,000 good parts per year for the shared cell and 110,400 for separate loading. Additional maintenance for two systems costs 1.2 million tenge more per year. Contribution margin is assumed to be 3,000 tenge per additional sold part.

At demand of 103,000 parts, both options cover the production program, so the extra capacity produces no revenue. The shared robot is financially stronger. At demand of 115,000 parts, the business can sell the full output increase from separate loading, and the additional investment returns quickly.

This example shows why you cannot choose by robot utilization percentage. The shared-cell manipulator is busy less than half the time, yet it can be the right investment when demand is limited. Under a different calendar, the same machine downtime turns into lost orders, and the more expensive independent loading wins.

Test the result's sensitivity to four values: demand, contribution margin, availability, and changeover duration. If a small change in any of them changes the decision, the project is near the financial boundary. In that case, sign the equipment contract only after a verification time study and order confirmation.

The decision must withstand the production calendar

A shared robot is usually chosen for long cutting cycles, infrequent part changes, limited demand, and the ability to stop both machines for service. Separate loading is stronger with short cycles, frequent changeovers, a high cost of downtime per hour, and confirmed need for extra capacity.

Document the decision as an assumptions table with an owner for every number. Production confirms cycles and changeovers, sales provides order volume, process engineering is responsible for quality and tooling, maintenance estimates recovery, and finance sets contribution margin and the discount rate. The discussion of the preferred architecture then becomes a verifiable calculation.

When selecting machines and preparing commissioning, provide EAST CNC with the actual timeline, part range, batch plan, and access requirements so the layout does not rely only on the CNC nameplate cycle.

Do not treat average utilization as a guarantee. The model must withstand a long changeover, simultaneous requests from two machines, failure of a shared gripper, and a month of weak demand. If the more expensive option keeps its advantage after these checks, its extra capacity truly has value. If that advantage disappears, one shared robot will recover the investment sooner.