Risk Assessment of Repetitive Tasks: A Comparative Analysis among Different Methods to Update the Maximum Frequency Allowed

This study analyzes the maximum frequency of movement allowed in the no-risk area at different levels of duty cycle (DC) and applied force for the following methods used for the assessment of DUE-WMSDs: SI, ACGIH(TLV), OCRA Index and RSI.

Regarding the SI, a safe threshold limit set at 6 was also used to determine the maximum frequency of movement, since the method identifies an area of uncertain risk ranging from 3 to 6 [24].

In the following section, a summary of the hypotheses made and some of the calculations from the previous study [22] are briefly reported in order to have an idea of the origin of some of the data used in the discussion.

2.1 Hypotheses made for the calculation of the frequency for ACGHI(TLV) and Ocra Index

In order to calculate the maximum frequency of movement (number of efforts per second in the net time of the cycle) that each method allows, in no-risk conditions varying the force and the duty cycle, a very easy task was considered involving only one postural variable such as the ulnar deviation of the wrist, e.g. when using a joystick lever. Both the push forward movements and the return to neutral position movements can be considered active i.e. employing force.

We started considering the method with the lowest number of variables in the evaluation algorithm, in order to fix the same working conditions for the other methods. In our case it is the ACGIH(TLV). The equation determining the action limit of risk (calculated from the two extreme values indicated in the ACGIH(TLV) graphic is as follows:

$N P F_{A L}=5.56-0.556 * H A L$   (1)

in which NPF_AL is the Normalized Peak Force, which in our case, for a better comparison of the methods, was made to coincide with the force value provided on the Borg Scale (from now on indicated by BS) [25].

To calculate the frequency the following formula was used:

$f=\left(\frac{H A L}{6.56 \ln D C-3.18 H A L}\right)^{\frac{1}{1.31}}$   (2)

in which DC is the duty cycle and f is the frequency of movement (the number of work movements per second considered in the net time of the cycle), derived from the formula proposed by Radwin et al. [26] for the calculation of HAL. From Eq. (1) the border line conditions (action limit) were obtained when varying the force. These are the following conditions: BS=1, HAL=8.20; BS=2, HAL=6.40; BS=3, HAL=4.60, BS=4, HAL=2.81; BS=5, HAL=1.01.

The values calculated for HAL were then inserted in Eq. (2) and different DC values were hypothesized (100%, 80%, 60%, 40%, and 20%) to find the values of frequency allowed for the no-risk area. These are reported in Table 1, assuming a 1-minute cycle time.

The SI method considers more parameters of evaluation compared to the ACGIH(TLV), so other conditions were set, and the same goes for the previous hypotheses without altering the results already obtained. Since the ACGIH(TLV) leaves to the expert assessor the possibility to modify the final result in case of awkward posture, and this was not done, a non-harmful posture was hypothesized. Regarding the rhythm (speed) of work (R), although it also depends on the frequency, it was possible to assign, in all the different hypotheses made, the coefficient R=1, since the frequency allowed for the no-risk conditions that derives from our calculations never exceeded 1.7 movements per second, which can be considered a “fair” work rhythm. The duration of a repetitive task is assumed to be between 4 and 8 hours. Applying the following formula, the parameter F was calculated:

$I * D * F * P * R * T=3 $ or 6   (3)

Parameter F in Eq. (3) derives from the number of exertions in one minute (F’) according to the following correlation identified by the SI method: F=0.5 if F’<4; F=1 if 4<F’<8; F=1.5 if 9<F’<14; F=2 if 15<F’<19; F=3 if F’ ≥20.In Eq. (3) the variables are those used by the SI method: I=intensity of exertion; D=duration of exertion; F=Efforts/minute; P=posture; R=speed of work; T=duration per day, considering, on the basis of the previous hypotheses, P=R=T=1.

It is important to note that, even though F’ represents “frequency of movements”, it must not be confused with the frequency f in Eq. (2). In fact, F’ refers to the average exertions calculated in a time-period of one minute starting from the number of exertions in the cycle time. This calculation also includes any breaks in work within the cycle (work time + rest time within the cycle). The frequency f considered in Eq. (2) refers to the number of exertions per second within the net-time of the cycle. For a comparison of the frequencies f and F’ the following correlation must be considered, valid for the hypothesis made in a 1-minute cycle time.

$f=\frac{\frac{F^{\prime}}{60}}{\frac{D C}{100}}=\frac{F^{\prime} * 100}{60 * D C}$    (4)

Since we are looking for the maximum frequency allowed in the no-risk area, we chose the highest value of F’ for each corresponding F.

Regarding the values to assign to the OCRA Index algorithm, the previously set working conditions were maintained, but further specifications were made, since the method requires a greater definition of the parameters inserted in the algorithm. A hypothesis of non-existence of complementary factors was made in order to avoid influencing the results already obtained using the other methods. The following conditions remain valid for the previous results as well since the working time is fixed, so far, varying from 4 to 8 hours:

• hypothesis (A): 4 hours of repeated work arranged as follows: 2 consecutive hours in the morning and 2 consecutive hours in the afternoon, with a 1- hour lunch break, and the remaining time in non-repetitive work;
• hypothesis (B): 6 hours of repeated work arranged as follows: 3 consecutive hours in the morning and 3 consecutive hours in the afternoon, with a 1-hour lunch break, and the remaining time in non-repetitive work;
• hypothesis (C): 8 hours of repeated work arranged as follows: 4 consecutive hours in the morning and 4 consecutive hours in the afternoon, with a 1-hour lunch break.

From the OCRA Index formula (mono-task job), in the respect of our hypotheses, the following relation is derived in the border line conditions of no-risk:

$\frac{A T A}{R T A}=\frac{F * T}{30 * a * b * c * d * T * e * g}=2$    (5)

in which a=F_om(force multiplier), b=P_om (posture multiplier), c=R_em (stereotype multiplier), d=A_dm (complementary factors multiplier), e=R_cm (recovery time multiplier), and g=D_um (duration multiplier). Parameter F in the formula of the OCRA Index method represents the average number of technical actions (in our hypotheses coinciding with a single effort or movement) per minute in a complete cycle, including the breaks, therefore, in agreement with what was previously considered for the SI method, correlation (4) in Eq. (5) was used from which, by simplifying, the following formula for the calculation of frequency is obtained:

$f=\frac{110 * a * b * c * d * T * e * g}{T * D C}$    (6)

in which f=number of technical actions per second allowed as maximum frequency in the no-risk range in the set conditions.

For the details regarding the definition of each multiplier in Eq. (6) refer to the following publication (Antonucci [22]).

2.2 Hypotheses made for the calculation of the frequency for the RSI method

To be consistent with the data coming from the previous study, the same previous hypotheses were conserved in this case as well.

Regarding the RSI algorithm, since 10 is the borderline value for the no risk area, the formula for the calculation of the maximum frequency of movements in the no risk area is as follows:

$R S I=I M \cdot E M \cdot D M \cdot P M \cdot H M=10$   (7)

in which the multipliers are those indicated by the method: IM=intensity of exertion (force); EM=exertion per minute (frequency); DM=duration per exertion; PM=hand/wrist posture; HM=duration of task per day.

The RSI method provides two formulas for the calculation of the HM multiplier related to the number of hours worked. In our hypotheses, since the number of hours worked was set from 4 to 8, the following formula was used:

$H M=0.042 \cdot H+0.09 \cdot \log _{e}(H)+0.477$    (8)

where, H is the number of hours worked in a day. Also in this case 3 different scenarios are supposed, similar to what was previously set for the OCRA Index algorithm. In this case HM assumes the following values: HM=0.77 (for 4 hours); HM=0.89 (for 6 hours); HM=1 (for 8 hours).

Regarding the Intensity of exertion multiplier (IM), as done before, instead of using the percentage of maximum strength (%MVC), we use the Borg CR-10 perceived exertion scale (BS) [25]. In this case, as reported by the RSI method [23] for BS equal or lower than 4, we use the following formula:

$I M=30 \cdot I^{3}-15.6 \cdot I^{2}+13 \cdot I+0.4$  (9)

and for BS over than 4:

$I M=36 \cdot I^{3}-33.3 \cdot I^{2}+24.77 \cdot I-1.86$    (10)

in which I varies from 0 to 1 according to the variation of BS from 0 to 10.

The results are the following: IM=25.61 (for BS=10); IM=15.08 (for BS=8); IM=8.79 (for BS=6); IM=6.70 (for BS=5); IM=5.02 (for BS=4); IM=3.71 (for BS=3); IM=2.62 (for BS=2); IM=1.57 (for BS=1).

Regarding the hand/wrist posture multiplier (PM), a non-harmful posture was hypothesized in this case as well. In this hypothesis, the PM multiplier can be fixed as 1.

Regarding the calculation of the duration per exertion multiplier (DM), since we hypothesized a 1-minute cycle time, the following formula is used:

$D M=0.45+0.31 \cdot D$    (11)

in which D is the duration per exertion measured in seconds. The parameter D is calculated by dividing the total exertion time (in seconds) by the number of exertions done in the same time, in other words, it is the reciprocal of the frequency (f) that we are looking for. Therefore, to calculate DM the following formula can be used:

$D M=0.45+0.31 \cdot\left(\frac{1}{f}\right)$   (12)

in which f is the number of exertions per second referring to the net time of the cycle.

The efforts per minute multiplier (EM) depends on parameter E (efforts per minute) which is calculated by dividing the number of efforts in a cycle by the duration of the cycle (in minutes). This parameter E, can also refer to the frequency (f) that we are looking for. In fact, having hypothesized in our case a 1-minute cycle time, and assuming there is no rest time from one cycle to another except for the time within the cycle, in which there are no efforts determined by the duty cycle (DC), we can express E in terms of DC and f as the following formula:

$E=\frac{f \cdot 60 \cdot D C}{100}$   (13)

in which as specified before:

$f=\left(\frac{\text {number of exertions}}{\text {work time in the cycle (in seconds)}}\right)$

$D C=100 \cdot\left(\frac{\text { work time in the cycle }}{\text { total time of the cycle }}\right)$

Since the RSI method uses two different formulas for the calculation of the EM multiplier which depends on the number of efforts per minute, in the first step of our calculation both formulas are used. Then, looking at the results, the right one was chosen based on the value of frequency f.

As an example of the calculations done, in the hypothesis of a total work time of 4 hours a day, DC= 60% and force (BS) = 2, the formulas for the calculation of the maximum frequency of movements in the no risk area for the RSI are as follows:

$2.62 \cdot\left(0.1+0.25 \cdot\left(\frac{f \cdot 60 \cdot 60}{100}\right)\right) \cdot\left(0.45+0.31 \cdot\left(\frac{1}{f}\right)\right) \cdot 1 \cdot 0.77=10$   (14)

$2.62 \cdot\left(0.00334 \cdot\left(\frac{f \cdot 60 \cdot 60}{100}\right)^{1.96}\right) \cdot\left(0.45+0.31 \cdot\left(\frac{1}{f}\right)\right) \cdot 1 \cdot 0.77=10$   (15)

We choose the value of f deriving from Eq. (14) or Eq. (15) according to the value of E obtained using Eq. (13). If E is below or equal to 90 (efforts/min) the correct f value r is the one deriving from Eq. (14), otherwise, if E is over 90 (efforts/min), the correct f value will derive from Eq. (15).

This paper analyzed the maximum frequency allowed in the no-risk area at different levels of force and DC for some of the most used methods for the evaluation of DUE symptoms and/or musculoskeletal disorders (MSDs). The purpose of this paper is mainly to investigate the changes when the RSI algorithm is used instead of the previous version of the Strain Index (both when the safe threshold limit is set at 3 and when it is set at 6), comparing the results with those provided by the ACGHIH(TLV) and the OCRA Index algorithm. Even though the RSI algorithm introduces important modifications compared to the previous 1995 version - among those the most important is the use of continuous rather than categorical multipliers that allow one to solve the problem of having, in some cases, a jump in terms of results when there is little variation of the parameters evaluated - marked differences still remain in the comparison with other methods, when looking for the maximum frequency allowed in the no-risk area in the hypothesis of a 1-minute cycle time and good posture.

In fact, except for in the comparison between RSI-SI(3)-SI(6), the RSI provides similar results to the OCRA Index and ACGIH(TLV) in less than 1/3 of the overall combinations of force and duty cycle (including in the hypotheses that consider all results differing less than 0.3 efforts per second as equal, rather than approximating all frequencies less than 0.3 movements per second as zero).

What can be considered noteworthy, beyond the relative differences among the results, some of them normal when different algorithms are used, is that in some particular circumstances the methods seem to start from very different boundary conditions.

For example: in the hypothesis of a 1-minute cycle time, 8 hours of repetitive work and good posture:

• the ACGIH(TLV) method allows 6 exertions per second in the net time of the cycle, while the RSI less than 1 (0.86), in the hypothesis of BS=1 and DC=60%;
• in the same conditions, but when DC=100%, the ACGIH(TLV) method allows almost 2 exertions per second (1.7) while the RSI and the OCRA Index do not consider any frequency of movement as “safe” in this condition;
• in the same conditions as above, but when DC=20% and BS=5, the RSI method allows less than 0.5 (0.31) exertions per second, while using the OCRA Index it is possible to stay in the “safe” area even if the number of exertions per second is over 4 (4.68).

In light of the above considerations, what clearly emerged is that a greater amount of epidemiological data would be desirable in order to better define the boundary conditions that can be considered harmful for a worker performing repeated movements.

Some limitations of this study must be considered: (i) only the raw application of the algorithms was applied, no modifications by an expert assessor were made on the results; (ii) some of the results obtained in the calculation are probably rarely found in the real field; (iii) only a few combinations of duty cycle (DC) and force (BS) are considered in order to obtain the maximum frequency in the no-risk area; (iv) only a 1-minute cycle time was considered; (v) in the study, a hypothesis of a good posture was always assumed.