Quantifying Ergonomic Improvement and Operational Release After Deploying a Pneumatic Industrial Manipulator

00Executive Summary

Classify movements by the forces actually borne before selecting metrics for the original and modified process. For lifting and lowering in which a person supports the entire load with both hands, use RWL, LI and CLI from the NIOSH Revised Lifting Equation (RNLE). Once the manipulator supports the load, measure the operator's starting, braking, pushing, pulling, turning and positioning forces separately. Where the operator drives the horizontal axis with both hands and whole-body effort, use a calibrated force gauge, HSE RAPP and ISO 11228-2. Where a motor, cylinder or other power source drives the horizontal axis, RAPP does not apply; use three-axis hand-force measurement and a task-specific ergonomic assessment. Assess posture angles during exertion against the supplementary internal criterion in this article. Any conformity assessment under ISO 11226 or GB/T 42730 must satisfy the complete applicability conditions of the respective standard.

This report evaluates pre-deployment manual lifting and post-deployment assisted work separately. Manual-lifting release requires FILI and STLI≤1.00 for every fully manual subtask and CLI≤1.00 for the combined task. After deployment, manual load-bearing lifts during the normal cycle must equal 0. For a mixed workforce, uncertainty-corrected start/stop force must be no greater than 150 N in every cycle, and sustained force no greater than 70 N. When RAPP applies, all nine official factors must be green and the total score 0; otherwise, complete a task-specific assessment. The posture, data-integrity, and residual FMEA requirements applicable to each stage must also pass. Failure of any applicable criterion prevents release; averages, impressions, and percentage improvement cannot compensate for it.

In the fixed-input calculation example, the original process has two fully manual, two-handed lifting subtasks with different geometries and loads, at a combined frequency of 2 lifts/min, followed by 45 min of recovery after 45 min of work. The CLI, calculated without intermediate rounding, is 1.87690861858986, exceeding 1.000000, so the result is a fail. The post-deployment example assumes that the manipulator supports the entire load and the horizontal axis is unpowered. In the 50-cycle input matrix, the maximum adjusted starting/stopping force is 142 N and sustained force is 61 N; all nine RAPP factors are green/0, and manual lifting is 0 lifts. The calculated result passes the criteria in this article. This example may be described as “manual load-bearing lifting exposure eliminated by 100%, with the project pushing/pulling gate passed.” The post-deployment RNLE result must not be reported as LI=0. Actual project performance must be supported by the corresponding original records.

Key Conclusions

1. RNLE evaluates only fully manual two-handed lifting; it does not evaluate pushing, pulling, guidance, or positioning after deployment of the manipulator.
2. If any of average mass, maximum mass, H, V, D, A, frequency, duration, recovery time, or coupling classification is missing before deployment, the RNLE calculation is invalid and the workstation shall not be released.
3. After deployment, assess each cycle using its maximum resultant force plus U_res. Retain cycle-by-cycle results in the acceptance record; an average alone is insufficient.
4. For a mixed workforce, this study adopts the lower HSE guidance values as project criteria: start/stop force no greater than 150 N and sustained force no greater than 70 N.
5. Apply RAPP only to fully manual horizontal movement using both hands and steady whole-body effort while standing or walking. It does not apply to powered horizontal motion or upper-limb-only control. HSE does not define the total RAPP score as a universal action threshold. For applicable tasks, this study separately requires all nine factors green and a total score of 0.
6. Demonstrated improvement requires both elimination of manual load-bearing lifting and passing post-deployment force, posture, risk, and record checks. Improvement in one category cannot offset failure in another.

01Task Classification and Decision Logic

1.1 Model the Two Exposure Types Separately

Classify the motion from the actual force-sharing condition, not the equipment name. A normal-cycle segment is push/pull or guidance only when all three conditions are met: the manipulator or tooling supports the workpiece weight in a closed load path; the operator does not use the arms to keep the workpiece from falling; and the workpiece remains controlled after hand contact is removed. If any condition is not met, the segment is shared load-bearing or fully manual lifting. An RNLE task may be defined only for a discrete motion in which the device is disengaged and the operator supports the workpiece's full weight with both hands. When the device continues to provide partial support, record RNLE as not applicable and use three-axis hand-force measurement and a task-specific biomechanical assessment.

Determine RAPP applicability from the horizontal motion. Use Section A when horizontal drive output is 0 and the standing or walking operator uses both hands and smooth whole-body effort in front of the body. If drive output is not 0, or the operator is seated, one-handed, using only the upper limbs, or relying on external body support, record RAPP as not applicable. A score of 0 cannot replace a task-specific assessment.

1.2 All Applicable Criteria Must Pass

The individual conditions are combined using logical AND to determine release:

If any sub-gate is false, Release=Fail. Percentage improvement, satisfaction, production-rate gains or the performance of individual cycles cannot offset a failed criterion.

02Standards, Editions, and Application Rules

GB/T 31054-2014, Mechanical products—Computer-aided engineering—Terminology for finite element numerical computation, is a finite element terminology document and is not used for the manual handling assessment in this article. Manual lifting and carrying are analyzed under GB/T 31002.1-2014, within its stated scope.

This article distinguishes method sources from internal acceptance conditions. International and national standards define applicability and assessment frameworks; CDC/NIOSH provides the RNLE calculation; and HSE provides reference pushing/pulling forces and RAPP classifications. The 150 N and 70 N criteria, all-green/0 score, 50 cycles, 100 Hz sampling rate, resultant-force uncertainty, posture angles, FMEA thresholds, reassessment after changes and annual interval are acceptance conditions adopted for this study. The NIOSH, HSE, ISO and GB/T documents are not thereby interpreted as statutory occupational exposure limits in China. HSE values such as 150 N and 70 N are guideline figures, not weight limits or approved safe limits.

03Pre-Deployment Manual-Lifting Baseline: RNLE

3.1 Required Inputs

Record origin and destination geometry and task conditions separately for each lifting subtask. The following fields are required:

RNLE data are invalid if original records are incomplete, measuring equipment lacks valid calibration, origin and destination are not distinguished, or only a single posture is recorded. The task cannot receive a passing conclusion.

The input domain is also a hard gate. H shall be no greater than 63 cm; V shall be within 0–175 cm; D shall be no greater than 175 cm; and A shall be within 0–135°. Frequency shall lie within the official FM table for the selected V and duration category. Do not extrapolate a multiplier beyond any limit; record RNLE as not applicable, and do not release the motion through RNLE. Retain the raw mass record for every item and calculate L_avg and L_max directly from those records. Geometry uses the greatest exposure in valid cycles; nominal design dimensions do not replace measurement. If the destination involves regrasping, a brief hold, precision placement, or hand guidance, retain a complete calculation chain for origin and destination and use min(RWL_origin, RWL_destination). If all four are absent, record “destination calculation not required” with the supporting determination.

3.2 RWL and LI

When using centimeters and kilograms:

Obtain FM from the official NIOSH frequency table by V, frequency, duration, and recovery condition. Obtain CM from the official coupling table. Do not interpolate linearly or assign either multiplier by experience. When a destination calculation is required, calculate RWL at the origin and destination separately and use the smaller RWL, which produces the larger LI. Acceptance always uses unrounded results; two decimal places are for display only. Any unrounded project-gate metric greater than 1.000000 fails; exactly 1.000000 is permitted.

RNLE has explicit application limits: two hands fully support a controllable load while standing in the specified industrial environment. Do not apply the equation directly to one-handed lifting, seated or kneeling work, team handling, static holding, shared load-bearing between a person and equipment, carrying while walking, pushing/pulling, shoveling, vertical motion faster than 76 cm/s, a shoe–floor static coefficient of friction below 0.4, or an unstable load. Evaluate horizontal carrying distance, frequency, and cumulative mass separately under ISO 11228-1/GB/T 31002.1. Outside the limits, do not report an approximate LI; record RNLE as not applicable and perform a task-specific assessment.

04Multi-Task CLI and Input Integrity

Where a job includes two or more different loads, frequencies, geometries, coupling conditions, or destination heights, divide it into subtasks and calculate CLI. Record L_avg and L_max for each subtask. Set FM to 1 to calculate frequency-independent recommended weight FIRWL, then use L_max for FILI. Apply each subtask’s frequency to obtain STRWL and use L_avg for STLI. Rank subtasks by descending unrounded STLI; when STLI values are equal, rank the higher-frequency subtask first.

FM_(1...i) is the multiplier for cumulative frequency through subtask i. Look it up using the added subtask’s V category, duration, and recovery conditions. Do not substitute mean or maximum LI for CLI. Every subtask requires unrounded FILI≤1.000000 and STLI≤1.000000; the group requires unrounded CLI≤1.000000. If duration or frequency exceeds the official table, or cumulative frequency has no valid FM, RNLE release fails. A missing FM must not be treated as 1.

Apply independent checking to the data: the recorder is responsible for original measurements, the calculator for formulas and table lookups, and the independent reviewer for task decomposition and units. Records must include all three names, dates and version numbers. Lock spreadsheet formula cells and retain inputs, lookup sources and calculation results. Screenshots cannot replace the complete calculation file.

05Two-Handed Push/Pull and Guiding-Force Assessment

5.1 Motion Segments

Record each post-deployment cycle in six phases: approaching the handle, starting, steady guidance, turning/height adjustment, stopping/fine positioning and disengagement. Evaluate peak forces during starting and stopping, sustained forces during steady guidance, turning and height adjustment, and posture and unintended load support during approach and disengagement. Between clamping and release, cumulative time spent manually supporting the workpiece must be 0 s, and the number of manually load-bearing lifts must be 0.

5.2 Instruments and Calculations

Use a calibrated three-axis force gauge with at least ±500 N range per axis, resolution no coarser than 1 N, sampling at least 100 Hz, and calibration validity no longer than 12 months. Define Fx along travel, Fy horizontally across travel, and Fz vertically. Install the sensor in the hand-to-handle load path without changing handle height, grip spacing, or motion resistance. Zero before and after each test set. If any axis’s zero difference exceeds 2 N in absolute value, invalidate and repeat the entire set.

Calculate resultant expanded uncertainty U_res rather than directly adding single-axis certificate values. Divide each axis’s expanded uncertainty at k=2 by 2 to obtain u_x, u_y, and u_z, then include the standard uncertainties from zero offset, alignment, and repeatability:

Use the maximum calculated U_res within each decision window; it must not exceed 5 N. At Fres=0, set directional-derivative terms to 0 and retain zero, alignment, and repeatability contributions. If guidance lasts less than 1.0 s, use the mean of all window samples plus U_res,sus. The start window runs from the first Fres>10 N until speed remains within ±5% of its setting for 0.5 s. The stop window starts 1.0 s before arrival and ends after speed remains 0 for 0.5 s. Synchronize video, motion, and force signals within 0.1 s. Compare indicated force plus U_res with the threshold; never add uncertainty only after making the decision.

HSE guidance compares the minimum pushing or pulling force needed to start the load with its guideline figures. This article instead uses the maximum three-axis resultant force during normal operation plus U_res as an internal exposure criterion, which differs from the original HSE measurement method. The 150 N and 70 N values are numerical thresholds selected for this article, not statutory limits.

5.3 Test Coverage

For each design boundary condition, use 5 qualified operators from the actual job, each completing 10 consecutive full cycles, for 50 valid cycles total. The sample shall include the shortest and tallest individuals on the job roster; record the heights of all 5 operators in whole centimeters. If women work in the job, at least 2 of the 5 operators shall be women. Boundary conditions shall cover at least maximum rated workpiece mass, maximum working radius, minimum and maximum permitted handle heights, both travel directions, and the lowest permitted air-supply pressure or power state. Complete 50 cycles independently for each boundary condition; do not combine conditions.

Every valid cycle must satisfy:

The 150 N and 70 N thresholds use HSE guidance values for women and apply here to a mixed workforce. Any valid cycle exceeding either threshold fails. Averaging, P95, or excluding the operator who exceeded a limit cannot change that result. The one-sided 95% binomial failure-rate upper bound for 50/50 cycles with no failures is 1−0.05^(1/50)=5.82%.

06RAPP, Routes, and Posture Assessment

First determine whether RAPP applies. Use RAPP Section A when the horizontal axis has no powered output and the operator, standing or walking and using both hands and whole-body effort, moves overhead-rail or wheeled equipment smoothly. If the horizontal axis has powered output, the motion is performed only by the upper limbs, operation is one-handed or seated, or the person relies on external body support, record RAPP as not applicable and use three-axis hand-force measurement plus a task-specific ergonomic assessment. The work shall not be released until that assessment is complete; a RAPP score of 0 cannot substitute for it.

For movements within the scope of RAPP, use the official color classifications: green indicates low risk, amber medium risk, red high or very high risk, and purple unacceptable risk. The HSE total score supports comparison and prioritization; it is not a universal action threshold. This article additionally requires all nine factors, A-1 through A-9, to be green and individually scored 0, with a total score of 0. Any amber, red or purple factor results in failure.

Forward flexion 20°, rotation 10°, wrist flexion/extension 15°, and radial/ulnar deviation 10° are internal supplementary gates in this study. Evaluate force-phase angles against these gates; assess conformity with ISO 11226 and GB/T 42730 only under each standard's complete applicability conditions. Use the maximum absolute angle over the full cycle, not a cycle-average angle. Equipment rated capacity is a separate hard gate: maximum workpiece mass shall not exceed the lowest rating among the equipment, end-effector tooling, lifting points, and connectors. When RAPP applies, exceeding an equipment rating is purple/unacceptable and requires an immediate stop.

07Test Design, Data Quality, and Traceability

The release package must contain every record below. Any omission fails the data criterion:

1. Measured mass of every workpiece, L_avg, L_max, center-of-mass range, gripping points, workstation cycle time, and shift duration;
2. Raw H, V, D, A, F, duration, recovery, destination-control decision, and coupling classification for each pre-deployment subtask;
3. Rated capacities and allowable pressure/voltage ranges of the manipulator, tooling, lifting points and connections;
4. Force-gauge serial number, calibration certificate, u_x/u_y/u_z, u_zero, u_align, u_repeat, U_res, sampling rate, installation photographs, and zero records;
5. Anonymized operator ID, height, job qualification, and test conditions;
6. Original three-axis time histories, segment markers, and unrounded Fstart,c and Fsus,c for every cycle, alongside the summary;
7. Synchronized full-cycle video, RAPP applicability decision, applicable A-1 to A-9 records, task-specific assessment, posture angles, and route photographs;
8. RNLE worksheets, FMEA, deviation records, review signatures, software/form revisions, and final decision.

Only instrument disconnection, synchronization failure or a clearly identified external interruption unrelated to the process may invalidate a cycle, and the reason must be recorded during the test. Excessive force, difficult movement and poor posture must not be used as reasons for exclusion. Repeat invalid cycles until 50 valid cycles have been obtained for every boundary condition, and retain all excluded original files.

Fix equipment warm-up and test conditions: complete startup checks as specified by the manufacturer; record air pressure, voltage, lubrication, braking, and load state; use the maximum workpiece mass in the design envelope; and do not increase assistance settings before formal testing. Record parameter values, software version, balance setting, and speed setting in the configuration baseline. Export them again after the test for comparison; any difference invalidates the test set.

08Worked Pre-Deployment Lifting Calculation

This section demonstrates the baseline calculation with fixed inputs. The two fully manual, two-handed lifting subtasks differ in load, H, V, D, A and coupling classification. Each occurs at 1 lift/min, for a combined frequency of 2 lifts/min. Continuous lifting for 45 min is followed by 45 min of seated monitoring, giving recovery time/work time=1.0, with no further lifting afterward; the short-duration FM is therefore used. Neither task requires regrasping, holding, precise placement or hand guidance at the destination, so only the origin is calculated.

8.1 Multipliers and Single-Task Results

Calculate every multiplier from the exact input, never by back-calculating a rounded display value. Use full spreadsheet precision for FIRWL, FILI, STRWL, STLI, and CLI decisions. Descending unrounded STLI places A first and B second. A exceeds 1.000000 for both FILI and STLI and fails; B passes both.

8.2 Composite Lifting Index

Adding B gives a cumulative frequency of 2 lifts/min. Look up the incremental FM using B’s V category: for short duration and V≥75 cm, previous frequency ΣF_A=1 lift/min gives FM_B(ΣF_A)=0.94; combined frequency Σ(F_A+F_B)=2 lifts/min gives FM_B[Σ(F_A+F_B)]=0.91:

The unrounded baseline CLI is 1.87690861858986, and subtask A exceeds both the FILI and STLI limits. The input condition therefore fails release for manual lifting. A load of 18 kg being below 23 kg does not establish a pass, because 23 kg is the load constant. After accounting for geometry, asymmetry, frequency and coupling, the STRWL for subtask A is 9.76811733625 kg.

09Worked Post-Deployment Push/Pull and Guiding Calculation

Workpiece mass remains 18 kg. Throughout the normal cycle, the manipulator and tooling independently support the workpiece weight; manual support time is 0 s and manual load-bearing lifting frequency is 0 lifts/min. The equipment supports the vertical axis, and horizontal-axis drive output is fixed at 0. The operator stands or walks, keeps both hands on the handle, and uses whole-body effort for starting, 2 m of horizontal guidance, steering, stopping, and positioning. This motion therefore meets the RAPP Section A applicability conditions in this report; RNLE is not calculated.

This section constructs calculation inputs from each person's maximum across 10 cycles. U_res is set to 3 N for every result, and the adjusted value is the input maximum plus 3 N. The following example demonstrates data aggregation and evaluation.

Across the assumed 50 cycles, corrected maxima are 142 N and 61 N:

Use the nine official Section A factors in the example. The moving overhead-rail assembly and tooling total 22 kg; adding the 18 kg workpiece gives 40 kg moving mass, within the equipment’s permitted 60 kg.

With the fixed inputs and 50-cycle summary matrix in this section, the calculation passes every project gate. Zero failures in 50/50 illustrates project sampling acceptance; interpret the population failure rate using the statistical upper bound in Section 5.3. It does not demonstrate that an actual project has completed validation. The improvement in this example is “manual load-bearing lifting frequency reduced from 2 lifts/min to 0 lifts/min, an elimination rate of 100%; the post-deployment pushing/pulling force and RAPP project gates passed.” Do not report “CLI reduced from 1.87690861858986 to 0” or “LI improved by 100%,” because RNLE is not applicable after deployment.

10Improvement Metrics, Release Criteria, and Decision Statements

10.1 Quantitative Metrics

Report the following improvement metrics side by side, retaining each metric's units and evaluation basis:

If discrete fully manual lifts remain after deployment, recalculate FILI, STLI, and CLI for those tasks and report the residual lifting frequency. Elimination is then below 100%, so do not claim complete elimination. For shared load-bearing, use three-axis hand-force measurement and a task-specific biomechanical assessment, not RNLE. A force margin below 0 fails. A positive margin means the criterion passes; it is neither rated equipment capacity nor a safety factor.

10.2 Release Criteria — Every Applicable Item Must Pass

Issue the corresponding conclusion from the results of all sub-gates:

• Pass: “All G1–G9 criteria are satisfied for the specified operating conditions. The task is released within those conditions; changes beyond them require reassessment.”
• Fail: “G [number] is not satisfied. The task is not released. Complete corrective work and repeat the full test plan before applying for release.”

Record the final decision as pass or fail.

11Risk Analysis and FMEA

Score severity S, occurrence O, and detection D from 1 to 5 and calculate RPN=S×O×D. Residual RPN must be below 20. An S=5 failure requires two independent controls, one preventive and one detecting or isolating, even when residual RPN is below 20. Use the following fixed scoring definitions:

O or D may be reduced only when the control architecture, test records and review signatures are all available. If any evidence is missing, retain the initial values. The table below is a fixed-input FMEA calculation example showing how control measures relate to residual risk.

E1–E7 define the evidence required to reduce O/D. Use a listed residual score only after those records are complete and reviewed; otherwise use the initial score. FMEA, force measurements, and applicable RNLE calculations form the release evidence. After any control change, recalculate residual risk and repeat G1–G9; do not simply change the RPN number.

12Change Control, Periodic Review, and Operating Instructions

Any of the following changes or abnormalities immediately invalidates the existing release and requires G1–G9 to be repeated in full:

1. Workpiece mass exceeds the released maximum or its center of gravity leaves the approved three-dimensional envelope;
2. Radius, handle height, travel distance, speed, or acceleration leaves the tested minimum-to-maximum range;
3. Cycle frequency, continuous-work duration, recovery time, or shift pattern exceeds the RNLE/RAPP assessment inputs;
4. Changes to the manipulator, tooling, lifting points, handles, brakes, rails, hoses, cables, air pressure, voltage, speed, or control recipe;
5. New slopes, steps, corners, slippery conditions, or obstacles on the route;
6. A dropped load, collision, gripping failure, excessive start or sustained force, or ergonomic injury;
7. Invalid force-gauge calibration, lost raw data, or an identified calculation or task-decomposition error.

No percentage allowance or dimensional tolerance exempts an exceedance from reassessment. Exceeding the frozen range by 1 minimum recording unit invalidates release. Do not apply 5%, 50 mm, 100 mm, or 10% exemptions outside the approved range.

Even without changes, inspect equipment, tooling, brakes, rails, and routes before each shift. Annually repeat 50 cycles for the frozen condition producing the greatest Fstart,c or Fsus,c, and update RAPP, posture records, and FMEA. Count the interval from the formal release signature date. If retesting is overdue beyond 12 months, the task becomes not released.

The on-site standard work instruction shall state maximum workpiece mass, permitted working radius, handle-height range, permitted pressure/voltage, speed recipe, normal motion sequence, prohibition of manual lifting, abnormal-stop procedure, and pre-shift inspection items. A training record proves only that training occurred; it does not establish competence. An operator becomes qualified for the job only after completing 10 consecutive cycles under controlled conditions with every force and posture gate satisfied.

13Conclusions

Ergonomic assessment should follow the changes in movement before and after deployment. First establish RNLE baselines for RWL, FILI, STLI and CLI for fully manual, two-handed lifting, then check whether manual load support has been eliminated. Use three-axis hand-force measurement and a task-specific biomechanical assessment for shared load support. For pushing and pulling with the load fully supported by the manipulator, measure starting/stopping peaks and sustained force in every cycle, selecting RAPP or a task-specific assessment according to whether the horizontal axis is powered. Combine the force results with posture, rated capacity, data completeness and FMEA results to determine release.

Select criteria for the actual task. Fully manual two-handed lifting uses RNLE with unrounded FILI, STLI, and CLI≤1.000000. Fully assisted post-deployment work requires manual load-bearing frequency of 0 lifts/min. Every cycle must satisfy start/stop resultant force plus U_res≤150 N and sustained resultant force plus U_res≤70 N. Applicable RAPP factors must all be green with a total of 0; otherwise, a task-specific assessment must pass. Internal posture limits are forward flexion≤20°, rotation≤10°, wrist flexion/extension≤15°, and radial/ulnar deviation≤10°. Residual RPN must be <20 and the evidence package complete. Failure of any applicable criterion prevents release.

The fixed-input example gives the following results: before deployment, subtask A has FILI=1.73216592487162 and STLI=1.84272970731024, and the two-task CLI=1.87690861858986, giving a fail. The post-deployment inputs give a manual lifting elimination rate of 100%, a maximum adjusted starting/stopping force of 142 N, sustained force of 61 N, all nine RAPP factors green and a total score of 0, giving a pass against the listed criteria. These results apply only to the operating conditions and calculation inputs in this article. Interpret the population failure rate using the one-sided 95% upper bound in Section 5.3. Actual projects must issue a “Pass” or “Fail” conclusion from complete test and review records.

References

1. CDC/NIOSH. Revised NIOSH Lifting Equation. https://www.cdc.gov/niosh/ergonomics/about/rnle.html
2. CDC/NIOSH. Applications Manual for the Revised NIOSH Lifting Equation, DHHS (NIOSH) Publication No. 94-110, revised September 2021. https://www.cdc.gov/niosh/docs/94-110/default.html
3. Waters TR, Putz-Anderson V, Garg A. Applications Manual for the Revised NIOSH Lifting Equation. Cincinnati, OH: National Institute for Occupational Safety and Health; DHHS (NIOSH) Publication No. 94-110, revised September 2021. DOI: 10.26616/NIOSHPUB94110revised092021. https://stacks.cdc.gov/view/cdc/110725
4. Garg A, Waters T, Putz-Anderson V. Multiple-Task Analysis Using Revised NIOSH Equation for Manual Lifting. Advances in Industrial Ergonomics and Safety VI, 1994:67–70. CDC Stacks hosted copy. https://stacks.cdc.gov/view/cdc/191326/cdc_191326_DS1.pdf
5. UK Health and Safety Executive. Risk assessment of pushing and pulling. https://www.hse.gov.uk/msd/pushpull/index.htm
6. UK Health and Safety Executive. INDG478: Risk assessment of pushing and pulling (RAPP) tool. https://www.hse.gov.uk/pubns/indg478.htm
7. UK Health and Safety Executive. INDG478 RAPP tool, official PDF. https://www.hse.gov.uk/pubns/indg478.pdf
8. UK Health and Safety Executive. Assessing pushing and pulling risks, including guideline forces. https://www.hse.gov.uk/msd/pushpull/assessment.htm
9. ISO. ISO 11228-1:2021 Ergonomics — Manual handling — Part 1: Lifting, lowering and carrying. https://www.iso.org/standard/76820.html
10. ISO. ISO 11228-2:2007 Ergonomics — Manual handling — Part 2: Pushing and pulling. https://www.iso.org/standard/26521.html
11. ISO. ISO 11228-2:2007/Amd 1:2022. https://www.iso.org/standard/80954.html
12. ISO. ISO 11226:2000 Ergonomics — Evaluation of static working postures. https://www.iso.org/standard/25573.html
13. ISO. ISO/TR 12295:2014 Ergonomics — Application document for International Standards on manual handling and evaluation of static working postures. https://www.iso.org/standard/51309.html
14. State Administration for Market Regulation, National Standardization Administration Committee. GB/T 31002.1-2014 "Ergonomics - Manual Operation Part 1: Lifting and Transferring". https://openstd.samr.gov.cn/bzgk/std/newGbInfo?hcno=7C21BD4A0BD8998FFF1118EEC6B2F17D
15. National Standardization Administration. National Standards Program Notice, Plan 20256325-T-469 "Ergonomics Manual Operations Part 1: Lifting, Lowering and Transferring". https://std.samr.gov.cn/noc/search/nocPlanDetailed?id=4507276B92189390E06397BE0A0AD3B9
16. State Administration for Market Regulation, National Standardization Administration. GB/T 42730-2023 "Ergonomics - Static Working Posture Assessment". https://openstd.samr.gov.cn/bzgk/std/newGbInfo?hcno=56FED88EAF0FD6FA568584085949ED71
17. State Administration for Market Regulation, National Standardization Administration Committee. GB/T 36954-2018 "Machine Safety - Application of Ergonomics Principles in Risk Assessment and Risk Reduction". https://openstd.samr.gov.cn/bzgk/std/newGbInfo?hcno=5D9BBA21BFF65103898B5E84240A0A60
18. State Administration for Market Regulation, National Standardization Administration Committee. GB/T 31054-2014 "Terminology for Computer-Aided Engineering of Mechanical Products Finite Element Numerical Calculation". https://openstd.samr.gov.cn/bzgk/std/newGbInfo?hcno=F0BE9B45DFE8EA0528C0BCBD8D57065D