How do you measure kitchen staff before automation?

The first step in assessing automation is not to ask “how many people can be saved”, but to ask how many person-minutes have been spent on the same task, how much equipment has passed and whether the work has disappeared, shortened or transferred. Middleby's article suggests that the total operating and labour costs should be known before the automated investment is judged [1]. It gives a macro thought, but does not prove that a particular kitchen move can be replaced, nor does it provide savings for the shop.

Separate cost from time

Cost accounts are defined financially and may include basic wages, overtime, taxes or benefits borne by employers, outsourced labour and other actual costs. Inclusion must be consistent with exclusions, periods, documentation and rules for allocation. Wage information is only aggregated using posts or cost centres, limiting access and not replicating individual salaries on site.

Time track actions, tasks and parallel relationships. It uses anonymous employee codes, jobs, shifts, task codes, start and end points, locations, equipment identities, anomalies and sources of evidence. Observation is not used to withhold wages or rank individuals; the aim is to identify opportunities for stable duplication, waiting, return to work and handover in the system. The two accounts are linked only to the approved post rates and comparable periods and do not extend the scope of personal information in pursuit of “accurate costs for each person”.

Four time groups before counting minutes.

Active operations are actions that employees must perform, such as loading, setting, observation determination, unloading or cleaning in accordance with approved procedures. It must be guarded that the equipment or food is running, and the staff cannot leave because of safety, process or field rules. The waiting period for departure is when the equipment is in operation and the employee can legally and safely do another job. Non-allocation includes class buffers, pre-shift preparation, public handover or other defined categories that could not be returned to a given mission at the time.

The need to guard and the possibility of leaving cannot be judged subjectively by observers. The code shall be based on the same type of document, mission protocol, food and safety conditions and the valid requirements of those responsible on duty. In the event of a dispute, the waiting period is marked as pending and is not written in advance as saving labour.

Two examples of parallel and double.

The first calculation is a purely hypothetical collaboration mission. Staff members A and B worked on a trans-shipment and distribution mission for 10 minutes at the same time, and both were occupied. The duration of the mission is 10 minutes and the human minutes are 2 people x 10 minutes = 20 men. It is not possible to keep one person in mind because the clock on the wall is only 10 minutes away; nor is it possible to write down 20 minutes of equipment operated by both persons.

The second calculation is a purely hypothetical parallel task. The employee cc starts the device mission with 2 minutes X and the device runs itself for 8 minutes; during this period cc completes the mission with 6 minutes Y under permitted leave and then returns to X for 1 minute to unload. X's manual is 3 person-minutes and equipment is 8 minutes; Y's manual is 6 person-minutes. The nine actual working minutes of c could not be added to the full manual value of X, otherwise the same time period of the same person would be duplicated.

If c has to focus on the guards at the last 2 minutes of X running, these 2 minutes are classified as X and must not be allocated to Y at the same time. If Y is interrupted as a result, the interruption and restoration shall be recorded on the basis of actual time. An action may have a master task code and a context label, but a person-minute is classified in only one main category.

It's not like it's going to grow in infinite minutes.

For each observation session, the following constant equation is used: the lined-in person minutes = the target task is active + must be guarded + the other encoded tasks + the handing over and cleaning + is not allocated to the time on duty. Whether rest, leave, pre- or post-class is included, in a separate list of pre-determined financial and working calibres, and not imposed on an equipment task.

It is assumed that each of the two employees will have 450 net work minutes, for a total of 900 person minutes. If the task account is combined into 940, check first whether the parallel time period is double-counted, whether the mission boundary overlaps, whether the equipment is awaiting adult writing; if only 820, check whether the handover, search, unusual communication and unallocated time are not recorded. The variance should not be apportioned pro rata to each mission but should revert to the original time line.

Re-entry and root cause analysis to the main account

Mistakes, re-engineering, re-engineering, finding tools, waiting for approval, failure recovery and unusual cleaning should not be deleted as “unnormal samples”. They retain original time, mandate, scope of impact, known facts and assumptions of cause. Reasons are classified as training, process, information, equipment, materials, environment or implementation factors only when there is sufficient evidence.

The term “individuals who are not responsible without investigation” does not mean that the enforcement factor is excluded. If the record shows that clear and available documents and training are available and that the equipment, materials and workspace conditions are normal, the deviation can be an element of evidence. Conversely, if marking is difficult to read, tools are missing, procedures conflict or training does not cover the task, “staff inexperience” cannot be used instead of systematic analysis.

The baseline only determines the next small test.

Candidates may be ranked by frequency, person minutes, fluctuations, return to work, must be guarded, standardization and food or safety conditions. The ranking above only indicates that it is worth designing a small back-to-back test for the same mission, load and period, and does not mean that there has been a proven gain from automation. The minutes saved during the mission may be offset by new loading, programming, monitoring, cleaning, replenishment, maintenance or unusual handling.

Even if the net human per minute falls, it cannot be directly converted to a job reduction. There is a need to check whether the period of release is concentrated, whether it is available for other valuable tasks during the real working hours, and whether scheduling, regulation, safety and service levels allow for adjustments. Therefore, the correct conclusion of the baseline account is usually “what is the priority measure” rather than “the number of people who have been purchased”.

The sample covers the mission structure, not just a busy day.

Prior to observation, the mission family is listed in historical orders or production records, including different menu combinations, batches, switch frequency, tools or pots, raw material form, cleaning segments and low to medium-high loads. For each of the mission members, a representative class is selected to record the events planned, actually occurring and not covered. A Friday night peak can show congestion, but does not represent openings, filings, shifts, patches and cycle-level cleaning.

The labels for normal days, anomalies and training days are to be kept in the sample. Normal days are used for steady-state missions, irregular days are used to understand the labour demand for missing materials, equipment anomalies, procedural conflicts or handover failures, while training days are used to indicate a learning period burden. The three categories should not be commingled into an average, nor should the anomaly be deleted in its entirety and the claim of real labour be made.

Differences between operators are to be kept, but not displayed by personal name. If a skilled person uses a significantly lower time than a newcomer, check whether this is caused by training, familiar working space, hidden memory, tool placement or task selection. Processes performed only by skilled people on the basis of memory cannot serve as a pre-automatic baseline “for all”.

Recording rules freeze before first shift.

The observation manual should provide operational definitions of mission start-up, termination, suspension, recovery, parallel, help, handover and return to work. For example, “waiting” cannot be marked by an observer only if the employee does not touch the device; the employee may make an observation judgement, record or maintain a security guard. The manual should describe the means of searching and allow field personnel to object to the code.

Prior to formal collection, the two observers could code the same authorized process independently, comparing their differences in mission boundaries, parallel relationships and time classification. When the difference is greater, the definition and the form are modified and the programme is not immediately ranked. Video or other means of recording are used only after obtaining the required authorization, a clear retention period, access rights and privacy boundaries, and are not justified by the need for accurate timing as an unlimited collection.

Device contact by action, not by function name

The contact sheet records what the employee actually did: remove the container, enter or choose the setting, observe and make decisions, transfer the material, clean, change the tools, request support or dispose of the abnormal. Tasks, inputs, settings, attachments and numbers of personnel are retained together. “Autonomy”, “intellectual” or “multiple” are functional descriptions only and cannot replace measurements of loading, guarding, unloading, cleaning and restoration.

The device is always kept in different fields for running time and manual time. The equipment program has been in operation for 15 minutes and only indicates how long the equipment has passed; how many of the operators are active in it, must be guarded or can wait to leave, which must be confirmed by synchronized time lines. This distinction is the basis for whether a follow-up small test can account for changes in work.

References

• Middleby: How to Determine Restorant Labor Costas and Saving Using Automation