Temperature Limit Values For Touching Cold Surfaces with the Fingertip (2024)

,

Article history

Abstract

INTRODUCTION

Contact between bare skin and cold surfaces may occur in industrial settings as well as in everyday activities. This contact may be intentional or accidental and can affect small skin areas such as a fingertip or larger areas such as the palm of the hand. The contact of human skin with a cold surface can cause pain, numbness and risk of skin damage (Daniels, 1956; Havenith et al., 1992; Chen et al., 1994a). Furthermore, the reduction in tactile sensitivity (Provins and Morton, 1960) and manual dexterity (Geng et al., 2000a; Geng 2001; Powell et al., 2002) associated with skin cooling also greatly influences one's ability to perform the manual tasks routinely required by workers in cold environments in an efficient and safe manner (Enander, 1984; Havenith et al., 1995; Heus et al., 1995). Despite the apparent health and safety risk owing to cold contact exposure, and a standard being available to determine temperature limits of hot surfaces in working environments (European standard, 1994, 1999), no such standard was available for cold surfaces.

The European Directive on Safety of Machinery requires that touching and handling of cold objects should be safe. Therefore the European Union funded COLDSURF research project was established in 1998 to collect data for skin contact with cold surfaces, as the available information in existing literature was sparse or unavailable (Holmér et al., 2003). A first report has been published reporting temperature limits for surfaces involved with long-term (up to 1200 s) gripping contact (Malchaire et al., 2002). The present study reports the results from the second part of the project concerned with the short-term (<120 s) and possibly accidental contact of cold surfaces with a smaller area of the skin, such as the fingertip. the results from both studies have been integrated in a proposal for an international standard for skin contact with cold surfaces (ISO, 2005).

For the purpose of assessing the cold contact response of the skin–material interface under typical conditions, several principal factors must be considered so that the environmental conditions are appropriately characterized. Surface temperature of contactable solid objects in industry varies greatly and may be as low as −30°C or as high as +10°C (Holmer et al., 2000). Furthermore the thermal properties of the contact material will extend from a low level of conductivity (e.g. wood) to highly conductive surfaces, such as metals. Indeed it has been previously reported that metal surfaces, in particular aluminium, present a risk of frostbite within 10 s at surface temperatures below −5°C (Chen et al., 1994a) during three-fingered contact. Another factor to be considered is the level of contact force which has been reported to influence skin cooling rate (Chen, 1994b; Geng et al., 2000b). Forces typically present may be as low as 1 N (fine motor tasks, or accidental contact) up to 10 N (accidentally falling without recovery) (Holmer et al., 2000).

Hand dysfunction and the associated safety risk during occupational practices in the cold increases with decreasing skin temperature. Onset of cold pain has been reported to occur between 23 and 14°C during cold contact (Havenith et al., 1992). A marked deterioration in tactile discrimination occurs at finger skin temperatures <8°C with numbness found in one-third of subjects at 7°C (Morton and Provins, 1960). The estimated freezing point of human finger skin is −0.6°C (Keatinge and Evans, 1960), but owing to supercooling it may be lower before freezing starts. As the skin surface temperature falls from −4.8 to −7.8°C, the risk of frostbite increases from 5 to 95% (Danielsson, 1996). It is therefore pertinent to investigate the relative risk of different levels of skin cooling that affect the safety of the worker.

The aim of the present study was to develop a predictive model to establish the temperature limits of cold touchable surfaces under conditions representative of those encountered daily in many occupations. This was achieved by (i) collating data to characterize index fingertip contact cooling across a broad range of materials, surface temperatures and contact forces; and (ii) using these data to empirically model the contact duration required for skin contact temperature to reach the physiological criteria of the onset of pain (15°C), numbness (7°C) and frostbite risk (0°C).

METHODS

The experiments were carried out with human subjects, whose distal phalanx of the index finger of the non-dominant hand touched a cold surface. The experiments were repeated for each subject under different experimental conditions [i.e. type of material, surface temperature of the materials (T_S) and contact force] in a balanced order. The response variable was contact temperature (T_C) measured over time.

Apparatus

Four materials, i.e. aluminium, steel, nylon-6 and wood, were chosen to be representative of a broad range of thermal properties (contact factor). The contact factor (F_c) is defined as the square root of the product of the thermal conductivity λ (W mK⁻¹), the density ρ (kg m⁻³) and the specific heat C [J (kg K)⁻¹]. The thermal characteristics of the materials are given in Table 1.

Solid blocks of the four materials with dimensions 96 × 96 × 96 mm were selected as the contact objects. The surfaces of materials were polished. The blocks were suspended on a counterbalanced system in order to regulate the finger contact force at 1.0, 2.9 and 9.8 N. The surface temperature of the contact material was achieved allowing the blocks to equilibrate with the ambient temperature (in either a cool-box or climatic chamber) before each exposure.

Data were collected at four different laboratories using either a hand-cooling box with only the hand exposed to cold air, or in a climatic chamber with the subject's whole body exposed to the cold. Hand-cooling boxes were located in climate controlled room (T_a = 20°C, RH = 50%) and were employed in two laboratories. Air temperature inside the cool-box was regulated between −35°C and +5°C using a temperature control module allowing an accuracy of ±0.5°C. Each cool-box had an access hole to allow the subject to insert their hand and part of the forearm. Also, the boxes were furnished with a three-layer Plexiglas window and lighting so that the hand could be observed. In the other two laboratories, experiments were performed in a climatic chamber regulated at an air temperature between −40 and +2°C and a relative humidity of 50%.

The contact temperature (T_C,°C) was measured continuously with copper–constantan thermocouples T-type of diameter 0.2 mm or copper–constantan thermocouples, 0.5 mm. The time constant of the sensor was 0.47 s and the accuracy was ±0.2°C (−5 to +36°C). The thermocouple was positioned in the middle of first phalanx of the non-dominant index fingertip and was secured by a small piece of surgical tape (Blenderm, 3M) leaving the tip of the thermocouple exposed.

Participants

After the experimental protocol was approved by the ethical advisory committee of the institutions conducting experimental trials, 40 healthy participants (20 women and 20 men) aged 18–35 years volunteered for the experimental study. They were not acclimatized to cold and none had any history of vascular or neurological disorders or cold injury. The instruction and information of the details, discomforts and risks associated with the experimental protocol were given to the participants before they signed a written consent. They were instructed not to consume alcohol 24 h before, drink tea or coffee or work physically during 1 h before the experiment. All subjects were right-handed, non-smokers and had experience with moderate cold exposure.

Anthropometric measurements of the index finger on the non-dominant hand of the subjects were carried out. The volume of the finger was measured by water displacement, immersing the index finger to first phalanx/proximal flexion crease of the finger. Finger contact area was measured by scanning a fingerprint at the three levels of contact force (1.0, 2.9 and 9.8 N) into a customized computer program. Each measurement was carried out three times and the average of the three measurements was used. Mean characteristics are shown in Table 2.

Procedure

Experiments were distributed across the four laboratories collecting data (10 participants per laboratory) in 24 experimental conditions in order to cover the broadest range of exposure in terms of surface temperature (T_S°C), and to replicate conditions with participants from different laboratories.

Each participant was tested over five sessions on separate days and at the same time of the day. Each experimental session began with the participant resting for a period of 30 min in a room, and the participant was asked to rate his or her whole-body thermal sensation on a thermal sensation scale (ISO, 1994). The sensation scale used was a reduced five-point bi-polar scale: +2 Warm, +1 Slightly Warm, 0 Neutral, −1 Slightly Cool, −2 Cool. Subjective votes were recorded at 5 min intervals. For the experiments carried out with a cool-box, the environmental conditions within the room were selected to induce the thermal sensation equivalent to −1 (slightly cool) at a standardized clothing insulation of 0.4–0.5 clo. For the experiments carried out in a climatic chamber, clothing insulation was selected to also achieve a sensation vote of −1. These levels of clothing insulation were dependent upon the environmental conditions of the chamber. For chamber conditions between +2 and −10°C, a clo value of ∼1.5 was worn, and for conditions between −15 and −40°C, a clo value of ∼2.0 was worn. The thermal sensation of slightly cool was targeted to achieve a state of vasoconstriction.

After the 30 min rest period, each participant touched, with the first phalanx of the index finger of the non-dominant hand, the appropriate material block at the required surface temperature with a given finger contact force. The skin cooling behaviour was monitored throughout the exposure until one of the following withdrawal criteria were met: a contact temperature (T_C) of 0°C; a test duration of 120 s; a typical sensation of frostnip about which the participants were instructed (burning/tingling); a sensation of intolerable pain or any other reason for which the participant perceived withdrawal to be necessary.

Database management

The experimental data collected from 1655 tests under 24 experimental conditions were stored and managed in a database. Afterwards a statistical analysis of the pooled database was conducted with the computer software Statgraphics Plus for Windows.

The following four parameters were extracted from the database:

• Duration: total time for the finger touching (D_SP, in seconds);

• Pain threshold: time for contact temperature to reach 15°C, [t(15), in seconds];

• Numbness threshold: time for contact temperature to reach 7°C [t(7), in seconds];

• Frostbite threshold: time for contact temperature to reach 0°C [t(0), in seconds].

The first parameter recorded was touching duration. The duration of touching was limited to 120 s. From the curve of evolution of the contact temperature during the experiment, the times needed to reach the three contact temperatures that are defined by t(15), t(7) and t(0), were determined by a mathematical interpolation or an extrapolation, the latter for maximally one extra point in the cooling curve. When the cooling curve was such that a certain contact temperature would never be reached, a missing value (—) was encoded, indicating that the data was absent. For instance, when the final T_C reached 10°C, the t(15) was interpolated, the t(7) was extrapolated and the t(0) was considered as missing value. For t(15) a missing value would imply that the touching time was longer than 2 min.

Statistical analysis

Statistical analyses were performed in all five laboratories participating in the project on their own data and subsequently on the merged dataset. The results indicated that most of the distributions of duration in a given condition were non-Gaussian and not symmetrical, especially for D_SP, limited to 120 s when the subject ended the 2 min exposure period. Therefore, the means and standard deviations were not computed and medians and quartiles were used for statistical analyses. As the number of data points for some conditions were rather low, the analysis of the limit values was based on the lower quartile in order to protect 75% of the population.

Empirical relations were derived based on the prediction of the lower quartile of the duration D_SP and the time to reach the contact temperature of 15, 7 and 0°C, respectively. The duration was empirically correlated with the surface temperature (T_S) and the contact factor (F_C) of the material (Table 1). A non-linear regression model was then derived to give the best prediction of the critical times [t(15), t(7) and t(0)] as a function of the surface temperature (T_S) and the contact factor (F_C). In order to make this prediction model as universal as possible, a single model was preferred over having different models for each condition. This choice could affect the predictive quality of the equation.

RESULTS

Descriptive statistics of the database

Descriptive statistics (mean, standard deviation, median, lower and upper quartiles) regarding spontaneous touching duration (D_SP) under 24 experimental conditions are given (Table 3). The number of tests for D_SP values <120 s are also given in Table 3. It can be seen that 752 of these 1655 duration were below the limit value of 120 s. The limit duration of 120 s was reached for most of the experimental conditions with wood and nylon. Also, the duration for touching aluminium and steel at surface temperatures higher than 0°C was 120 s (Table 3).

Descriptive statistics for the times to reach the contact temperatures (T_C) of 15, 7 and 0°C [i.e. t(15), t(7) and t(0)] are given (Table 4). The values could be evaluated in 1448 cases out of the 1521 for t(15), 1490 cases out of 1563 for t(7) and 990 cases out of 1130 for t(0), respectively. This is because of the number of missing values shown in Table 4.

Tables 3 and 4 show that the number of points available to determine the lower quartile (25th percentile, P₂₅) was limited under some conditions. This was dependent upon the combinations of thermal properties and surface temperature of the materials. Furthermore, the P₂₅ values for D_SP increase with increasing the material surface temperature (Table 3). However, the P₂₅ values for the times to reach the T_C of 15, 7 and 0°C display an inconsistent pattern for wood—the lowest contact factor (Table 4). This may reflect the distribution of the test conditions over the different laboratories and test environments, which for wood with its slow cooling would have the largest impact, as indeed was observed here.

Empirical modelling

• for time to reach T_C of 0°C, (R² = 0.99, only for steel and aluminium)

  $t(0)(980.6FC1.03)exp(0.21×TS)$

  (3)

• for time to reach 7°C, (R² = 0.99, only for nylon, steel and aluminium):

  $t(7)=(454.6FC1.801)exp(0.21×FC0.467×TS)$

  (4)

• for time to reach 15°C (R² = 0.93, only for nylon, steel and aluminium):

  $t(15)=(13.7FC1.09)exp(0.108×TS)$

  (5)

It is evident from Figures 1–3 that the time for the T_C to reach 15, 7 and 0°C are notably faster in the cases of touching at the lower surface temperature. The time to reach the critical temperatures when touching the cold metallic surfaces was significantly shorter than that when touching the non-metallic surfaces under all the conditions studied. The predicted values are lower than the observed values, indicating a degree of safety.

Since the P₂₅ values of time to reach 0°C from wood and nylon were longer than 100 s, the expression for the prediction of t(0) cannot be used for these non-metallic materials. The final prediction model for the time to reach the contact temperature of 0°C was derived only for steel and aluminium under conditions of T_S < 0°C in equation (3). Also, the P₂₅ values of time to reach 7°C from wood were longer than 100 s, which were not taken into account for the models of t(7) in equation (4). The data of P₂₅ of time to reach 15°C for material of wood were lower than 100 s (Table 4), but it was impossible to derive a valid prediction of the inconsistent patterns (Figure 3). The model for t(15) was derived from the data of P₂₅ from nylon, steel and aluminium in equation (5).

After incorporating the effects of material contact factor and surface temperature for the prediction of t(0), t(7) and t(15), it was found that variation in finger contact force did not contribute any further significant predictive power to the empirical model for any level of contact cooling risk.

Secure time thresholds

As mentioned earlier the results have proved that the surface temperature was a function of time for T_C to reach a criterion. The criteria were three levels corresponding to frostbite (T_C = 0°C), numbness (T_C = 7°C) and pain (T_C = 15°C). The thresholds for finger touching various cold surfaces are shown in Figures 4–6.

The experimental data did not show any occurrence of cooling to 0°C for the case of a finger touching nylon and wood. Hence the frostbite threshold for the contact with nylon and wood does not exist in the range T_S from −40 to 0°C (Figure 4). The thresholds for numbness and pain were <100 s when the finger touched these four materials at surface temperatures between −40 and +2°C.

DISCUSSION

The present study investigated the influences of thermal properties of a contact material, surface temperature and finger contact force upon skin cooling behaviour of the fingertip during short-term (<120 s) contact. The two principle factors influencing contact cooling time across the conditions and participants studied were thermal properties of the material (expressed as a contact factor) and surface temperature.

The aim of the model derived was to provide a universal equation that could be applied to a range of conditions to predict the contact duration required to reach skin temperatures associated with three different levels of contact cooling risk—all influencing safety in the workplace. The three levels of risk were the onset of cold pain (T_C = 15°C), the onset of numbness (T_C = 7°C) and the onset of skin freezing (T_C = 0°C). The final model proposed was a non-linear function of material contact factor and surface temperature, incorporating different constants for each predicted level of risk.

Greater contact forces have been previously demonstrated to significantly affect skin cooling behaviour of the contact site during exposure to metallic surfaces (Chen et al., 1994b), though no such effect was found for non-metallic surfaces (Geng et al., 2000b). In the model presented here, finger contact force was not found to add a significant amount of predictive power and was therefore omitted from the equation. The absence of significance for this effect may be attributed to data being collected in different laboratories, as well as to the mixing of whole body exposures and hand-arm-only exposures. These factors will have introduced more variability in the data and created more overlap between the contact force condition data. The predictive model in the present study was required to be universally applicable to the broad range of conditions typically present in the workplace, including both metallic and non-metallic materials and for a mix of contact force situations. Hence, it was deemed preferable to pool all obtained data into an as large as possible dataset, rather than to try to discriminate between all possible effects in more controlled but very small datasets separately. The data and equations presented are therefore representative of a wide range of contact cooling exposure types, and require fewer caveats than when only a small dataset would be analysed in more depth.

A further reason for the lack of significance of contact force differences may have been the level of vasoconstriction present in the participants who were ‘slightly cool’. In support of this notion, it has been demonstrated that a significant difference in contact cooling response exists between an arterial occlusion and a vasodilated condition (Jay and Havenith, 2005) showing the impact of blood flow on cooling responses. Furthermore, it has also been shown that this blood flow is modulated by the contact force, in that laser-Doppler skin blood flow of the fingertip reduces exponentially with increasing finger contact force, and finally, that the extent of this reduction decreases dramatically with a cooler thermal state of the lower arm (Jay and White, 2006). It is reasoned that where vasoconstriction is present due to a ‘slightly cool’ thermal state of the person, like in the present experiment, the closure of the capillaries by greater contact forces will have less impact than in a warmer state with high ‘uncompressed’ finger tissue blood flows.

The present study was conducted to develop limit values applicable to the general population; therefore the experimental study was performed upon both males and females as a group. The proposed models for the varying levels of contact cooling risk cover both males and females. Further studies have scrutinized the finger contact cooling data with respect to gender. A gender difference in finger cooling time has been reported, with typical females showing faster skin cooling rates than typical males under slow cooling conditions (Geng 2001; Jay and Havenith, 2004a). These differences have been attributed to typical differences in hand size between males and females, in terms of surface area and mass. Also, females have less capacity for metabolic heat production of the extremities (Pathak and Charron, 1987). In a subsequent study with a greater subject pool and by design matching gender for hand anthropometry (Jay and Havenith, 2004c), it was demonstrated that after uncoupling hand size and gender, no difference exists in contact cooling between males and females under slow cooling conditions. They also concluded that skin cooling was dependent upon hand/finger size of the individual, with the best predictor being finger volume (Jay and Havenith, 2004c). For a general prediction model as developed here, males and females were therefore not separated, but seen as representing the range of hand sizes present in the general population, independent of gender.

The effect of individual variability upon contact cooling response has been investigated (Rissanen et al., 2000; Geng, 2001; Jay and Havenith, 2004b). Individual tissue properties of finger skin, blood flow through the micro-circulation under the skin and subsequent heat input may also influence contact cooling behaviour (Holmér, 1997). Furthermore, initial skin temperature of the hand and metabolic rate of whole body as well as constitution of the contact were also considered. To cover most of the individual variation and ensure protection for 75% of the population in contact with cold surfaces, the non-linear predictive models are based on the database of lower quartile (P₂₅) values. These experiments were carried out with healthy human subjects between the ages 18 and 35 years, however it should be considered how these responses apply to the wider population, particularly as a function of age. Collagen fibre density of the human dermis increases with age up to 40 years, following which it begins decreasing at a similar rate. Furthermore, in the reticular dermis the relative volumes of elastic fibres also increase with age attaining the highest values beyond the 40s (Vitellaro-Zuccarello et al., 1994). These data suggest that the relative insulation provided by the skin of the fingertip should be similar at 60 years to that at 20 years. While it is acknowledged that other factors such as differences in peripheral blood flow may also change as a function of age, the models in the present study are considered to be applicable to the finger skin of healthy males and females between the ages of 15 and 60 years.

Contact temperature (T_C) was used as a measure of skin cooling response in the present study. Upon contact, equilibrium is achieved at some point intermediate between the two initial temperatures of the contacting skin and surface, this is the ‘contact temperature’ (Vendrik and Vos, 1957). It has been demonstrated that these relative values are reproducible and indicative of the absolute skin temperature values (Stoll, 1977), and they are used frequently by clinicians.

From an ergonomic point of view, an estimate of the skin cooling risk is possible by measuring the surface temperature of the cold object and the contact time to reach a defined criterion. As mentioned, earlier studies justify that the criteria would be levels corresponding to 0°C for frostnip (Danielsson, 1996), 7°C for numbness (Provins and Morton, 1960) and 15°C for pain (Havenith, et al., 1992; Geng, et al., 2001). The determination of contact time is more convenient than the measurement of the contact temperature. It is apparent that frostbite injury may occur with finger contact of an aluminium surface at −20°C for only 1 s, at −15°C for <3 s, and at −10°C for ∼5 s (Figure 4). The time thresholds are useful and informative for the protection of finger/hand skin in cold operations. They give recommended, longest contact times for different cooling criteria for design of workstations and hand tools.

The data obtained from two different sets of cold exposures (whole-body and hand-only) were intentionally mixed into the database. The first one refers to situations such as cold rooms where heavily clad individuals must touch cold and dry objects. The experiments did not concern conditions where the objects are wet or iced, however thin ice layers have been subsequently demonstrated to accelerate contact cooling rate (Rissanen and Rintamäki, 2005). The other exposure conditions referred to environments where the hands only are exposed to cold goods. Such conditions are likely to occur in industry where workers may contact refrigerated pipes or pieces of machinery. In some cases, these cold surfaces might be iced or wet, however these situations were not considered in the present research. In addition, fluids below 0°C may comprise an acute risk of frostbite. Examples of such fluids are ethereal gasoline, ammonia and nitrogen. Similarly, solid carbon dioxide gives rise to a risk of frostbite. Therefore it is necessary to consider that the risk for frostbite can increase during skin contact with cold liquid and other such substances.

CONCLUSIONS

In conclusion, the present data has demonstrated the strong link between the risk of skin damage and the thermal properties of the materials touched, with the aluminium falling into the highest risk group of the materials tested here. For aluminium and steel, skin temperatures of 0°C occurred within 15–20 s of contact at material temperatures as high as −4°C and within 2–6 s at −15°C. For non-metallic surfaces, onset of numbness occurred within 15–65 s of contact at −35°C and onset of cold pain occurred within 5 s of contact at −20°C. The predictive model subsequently developed was a non-linear expression incorporating the effects of material thermal properties and surface temperature. This new tool is used to protect workers against the risk of cold injury, numbness and pain by establishing the temperature limits of cold touchable surfaces for a broad range of materials, and it is now proposed as guidance values in a new international standard (ISO, 2005).

This study was supported by the European Commission project: Temperature limit values for cold touchable surfaces (EU Contract No. SMT4-CT97-2149).

REFERENCES

Author notes

^†Project work carried out at National Institute for Working Life, Solna, Sweden

© The Author 2006. Published by Oxford University Press on behalf of the British Occupational Hygiene Society

Issue Section:

Original Articles

Email alerts

Citing articles via

More from Oxford Academic