Thermal Work Environments – Part 10: The Search for a Universal Index

     Over the past century, many researchers have attempted to quantify the physiological impact of high- and low-temperature environments (see Part 4 and Part 8, respectively).  As knowledge of human biometeorology increased and computational tools became more powerful, the models used became much more sophisticated.  However, models are typically focused on one type of environment – hot or cold – requiring use of multiple indices to accommodate varying conditions.  Other shortcomings in thermal index formulations further limit their utility in highly-variable conditions.
     In this installment of the “Thermal Work Environments” series, indices presented earlier in the series are evaluated and compared.  Additional indices are also considered for recommendation in workplaces.  Finally, a universal index, valid across the foreseeable range of human environmental exposure, is presented.

Classification and Evaluation of Thermal Indices
     The classification and evaluation schemes summarized here are the work of C. R. de Freitas and E. A. Grigorieva (2015, 2017).  de Freitas and Grigorieva scoured scientific literature to catalog more than 160 thermal indices.  Characteristics of each index were used to sort them into eight categories, or classes, identified by letters A through H.  The following descriptions are used to differentiate the classes:
simulation device for integrated measurement;
  A:  single-parameter index;
  B:  index based on algebraic or statistical model;
  C:  proxy thermal strain index;
  D:  proxy thermal stress index;
  E:  energy balance strain index;
  F:  energy balance stress index;
  G:  special-purpose index.

     Each index was then scored on a 5-point scale for six evaluation criteria.  The criteria and scoring methods are summarized below.
     Comprehensiveness indicates the number of relevant variables accounted for in the index.  Any factor contributing to thermal stress or strain is a relevant variable, including air temperature, humidity, wind speed, insolation or other radiation exposure, metabolic rate, clothing and protective gear in use, etc.  Comprehensiveness is scored as follows:
   Each relevant variable = +1.
   Maximum score = 5.
     Scope indicates the range of conditions for which the index is valid.  Scope is scored as follows:
   A narrow range of conditions is covered = 1.
   A broad range of conditions is covered = 3.
   Both cool and cold conditions are covered = 4.
   Both hot and cold conditions are covered = 5.
     Sophistication indicates the theoretical soundness or empirical support of the index.  Sophistication is scored according to the classification scheme, described above, as follows:
   Classes A and B = 2.
   Class C = 3.
   Classes D and E = 4.
   Classes F and G = 5.
   Class H is scored according to the methods used (i.e. classes A to G, above).
     Transparency indicates the clarity and justification of the rationale underpinning the index.  Transparency is scored as follows:
   None of the terms used are justified = 0.
   Justifications for terms used are poor or weak = 1 or 2.
   Terms used are justified in most cases = 3.
   Terms used are justified = 5.
     Usability indicates the ease with which an index can be implemented and interpreted.  Evaluation is based on the presence or absence of three characteristics:

1. Computational procedures are straightforward.
2. Only ‘standard’ data are required.
3. Outputs are easily interpreted.

Usability is scored as follows:
   The index exhibits none of the three characteristics = 0.
   The index exhibits only one of the three characteristics = 1.
   The index exhibits two of the three characteristics = 3.
   The index exhibits all three of the characteristics = 5.
     Validity indicates the degree to which the index value accurately reflects the physiological impact, or human experience, of environmental conditions.  Validity is scored as follows:
   The index has not been validated = 0.
   A rational index that has not been validated = 2.
   The index has been compared to one fully-validated index = 3.
   The index has been compared to multiple fully-validated indices = 4.
   The index is derived from or tested with empirical data = 5.
     The six evaluation criteria are deemed to be of equal importance and are equally weighted in the final score.  The total score for each index, therefore, is simply the sum of the six criteria scores.  Using the de Freitas and Grigorieva scheme, scores can range from 4 to 30; indices with higher scores are expected to be more useful.
 
Heat and Cold Indices Revisited
     The merits of various indices used in hot (Part 4) and cold (Part 8) environments have been discussed.  The de Freitas and Grigorieva framework provides a consistent method of evaluation and comparison.  The summary table in Exhibit 1 presents the scores for heat and cold indices previously discussed.

     All of the heat indices listed in the summary table are direct indices, as discussed in Part 4.  However, identifying the “best” index for use in a hot environment is not as simple as finding the highest total score.  Although the scoring method weights each criterion equally, specific circumstances may influence practitioners’ preferences.  Also, scores may be disputed or obsolete, as the following examples demonstrate.
     Heat Index is given a comprehensiveness score of 5, though it could be argued that holding constant all but temperature (Tdb) and humidity (RH) reduces this to 2.  Updating its usability score of 3 may also be warranted.  De Freitas and Grigorieva (2017) state that “it is reasonable to argue that once user-friendly routines are made available (on a website, say) to run the calculations…, the usability would achieve a top score” in reference to another index.  This logic also applies to Heat Index, for which online calculators have long been available.
     Accepting both arguments changes the Heat Index score from 23 to 22.  While not a large change in total score, maintaining second rank on this list, the constituent scores give a different impression of this index.  Similarity of Heat Index and humidex may also inspire questions about the discrepancies in scores on these two criteria.
     Thermal Work Limit (TWL) provides another cautionary example.  It attained the highest score of the heat indices examined, scoring the maximum on four of the six criteria.  However, its narrow scope of applicability [36 – 40° C (96.8 - 104° F)] and data requirements (e.g. body dimensions) may preclude its use in some situations.
     Ranked third on our list by total score (20), Wet Bulb Globe Temperature (WBGT) is worthy of continued attention.  It is the basis on which ACGIH has set threshold limit values (TLVs) for heat stress.  The availability of instrumentation, simple calculations and estimation procedures seems to warrant a usability score of 5; instead, it has been scored a 3.  It should also be noted that WBGT’s comprehensiveness score of 3 applies to outdoor environments.  For indoor environments, dry bulb temperature (Tdb) is omitted, reducing the score to 2.

     Among the cold indices examined, The New Improved Wind Chill Index (Twc) and required clothing insulation (IREQ) tied with a high score of 26.  The tie-breaker, in occupational settings and other practical applications, is ease of use.  The simplicity of using Twc (usability = 5), whether calculated or estimated from a table, outweighs its deficiency in comprehensiveness by making consistent use more likely.
     The examples provided are not intended to be exhaustive or the conclusions to be absolute.  Reasonable people can disagree on individual scores or the method of scoring, particularly when reviewing a list as extensive as that compiled by de Freitas and Grigorieva.  It is important to recognize this potential for disagreement, understand its implications, and move past it to implement appropriate tools for one’s own situation.
 
Highest-Scoring Indices by Class
     To explore additional candidates for use in occupational settings, a list of the highest-scoring indices in each classification was compiled.  A summary table is provided in Exhibit 2, listing indices that tied for high score in each class.  This subset was then subjected to analysis, similar to that described above, to identify indices that exhibit sufficient potential for supplanting a familiar index to warrant in-depth investigation.  Admittedly superficial, the large number of indices necessitates an abbreviated analysis.  A description of this review, further condensed, follows.

     The leading index in Class A is Thermo-Integrator, despite only scoring 18 points.  Though fully validated (validity score = 5), its other scores are uninspiring; further research was eschewed.
     Class B maxed out at 14 points, where air temperature (Ta) and wet bulb temperature (Twb) tied.  Both are components of WBGT; use of either independently would be injudicious.
     The standout in Class C is Resultant Temperature (RT) or Net Effective Temperature (NET) with a total score of 23.  Its maximum scores for scope and usability make it an attractive option for consideration.  Unfortunately, an English-language version of the defining paper could not be found, ending the inquiry.
     Two indices tied for the top spot in Class D, with 24 points each.  The Index of Physiological Effect (IPhysE or Ep) assesses the level of strain on a points scale, while the Predicted Four-hour Sweat Rate (P4SR) uses liters of sweat produced as index values.  The impracticality of utilizing such indices in occupational settings was discussed in Part 4.
     Class E is topped by two indices with total scores of 25.  The defining paper for Classification of Weather in Moments (CWM) could not be found in English.  The description provided by de Freitas and Grigorieva of the output of this index is “weather types,” dampening enthusiasm for a continued search.  Effective Temperature (ET) is not valid in cold temperatures.
     There is a three-way tie, at 28, at the top of Class F.  Unfortunately, the three leading indices are impractical for use in occupational settings.  Body-atmosphere Energy Exchange Index (BIODEX) requires core temperature monitoring, while Skin Temperature Energy Balance Index (STEBIDEX) requires skin temperature monitoring.  The Subjective Temperature Index (STI) causes concern with its name alone.  It also requires “nonstandard” data and the maximum valid temperature is 40° C (104° F), which excludes settings that are in great need of effective monitoring and controls.
     Atop Class G is a four-way tie at 28 points.  Of the four high-scorers, only the Standard Effective Temperature for Outdoors (OUT_SET*) outputs an equivalent temperature, making it the most-easily understood index.  Its use of “nonstandard” data reduces its usability score and, thus, its practicality for use in occupational settings.  Its focus on outdoor environmental conditions also limits its applicability.
     Bucking the upward trend in high scores, Class H falls back to 26 with a tie between two indices with questionable value in our chosen context.  The Acclimitization Thermal Strain Index (ATSI) is focused on the physiological adaptations required for travel; investigation of the Bioclimatic Distance Index (BDI) was thwarted by another language barrier.  Maximum valid temperatures for both indices are also rather low.  A lack of viable candidates in the special-purpose category is not surprising.

     The search for a thermal index to solve all our problems, based on high scores in the de Freitas and Grigorieva framework, has been rife with disappointment.  The context of use in occupational settings, in all potential permutations, has heightened the challenge.  Changing focus slightly reveals a new subset of indices; these candidates are reviewed next.

Other Candidate Indices
     Widespread familiarity and accessibility of Heat Index (HI) and Wind Chill Temperature (Twc) charts and calculators and the resultant levels of effectiveness they have achieved sets a high bar for any potential replacement.  High total scores in the de Freitas and Grigorieva scheme is an ineffective metric for identifying potential replacements for the HI/Twc duo, as shown in the previous section.  Instead, focus must be narrowed to the characteristics that determine the feasibility of an index in the context of concern – any occupational setting.  In this context, two scoring categories stand out.
     To be considered universal, an index must be valid throughout the range of conditions that may be encountered.  In the de Freitas and Grigorieva scheme, this equates to a scope score of 5.  A single index should eliminate the gap in valid temperatures between Twc [-40 – 10° C (-40 – 50° F)] and HI [20 – 60° C (68 – 140° F)].
     To ensure consistent, reliable application in occupational settings, an index must be simple to implement.  Here, this equates to a usability score of 5.
     Sorting the list according to these criteria yields the set of indices shown in the summary table in Exhibit 3 (previously-reviewed indices are excluded).  These indices were subjected to analysis similar to that described above; a brief review follows.

     The first index listed, Effective Temperature (ETM) includes only two variables in its calculation.  Other indices, including the HI/Twc duo, are often criticized for a lack of comprehensiveness; it would be difficult to argue that popularizing a “new” index with the same shortcoming is worthwhile.
     The next seven indices on the list scored 0 for validity.  Without empirical support, none of these indices is likely to gain traction as a potential replacement for HI or Twc.  Without some level of validation, use of these indices is also too risky when worker well-being is at stake.
     The Class G indices in this list show promise, though most score low on validity; additional work is needed for them to be viable alternatives.  The exception is the Universal Thermal Climate Index (UTCI), which warrants a closer look.  An overview of UTCI is provided below.

The Universal Thermal Climate Index
     Development of the Universal Thermal Climate Index (UTCI) was initiated by the International Society of Biometeorology (ISB) and the European Union COST (Cooperation in Science and Technology) program’s Action 730.  Objectives of the index development project, which involved scientists from 23 countries, included:

• An index based on advanced thermophysiological models, incorporating all modes of heat transfer between the human body and the surrounding environment.
• An index capable of predicting whole-body effects (e.g. heat stroke) and local effects (e.g. frostbite).
• An index valid for all conditions and all “time and spatial scales.”
• An “apparent temperature” index, where the output is the air temperature of a reference environment that causes the same physiological response as the input conditions.
• An index that requires minimal computational capacity, allowing rapid, widespread application.

      A visualization of the UTCI calculation procedure is provided in Exhibit 4.  The generic mathematical formulation of the calculation is as follows:

where Ta is air temperature, Tr is radiant temperature, va is wind speed, and pa is vapor pressure (humidity).  Mean radiant temperature (Tmrt or MRT) is typically used for Tr and can be calculated as follows:

where Tg is globe temperature (°C), ε is emissivity of the globe, and D is the diameter of the globe (m).  For a standard globe, such as that typically used for WBGT measurements, where D = 0.150 m and ε = 0.95,

This is an approximation of a more-rigorous calculation accounting for direct, indirect, and reflected solar radiation and infrared radiation from the sky and surroundings.

     Meteorological data typically include wind speeds measured at a height of 10 m (33 ft).  Heat balance models often use a height of 1.1 m (3.6 ft) as the average height for human exposure to wind; other heights are needed in the detailed calculations of the clothing model (e.g. head, upper leg, etc.), discussed below.  Meteorological wind speeds can be converted to an appropriate height for specific calculations according to the following:

where va is wind speed (m/s) at height Z (m), vZr is wind speed (m/s) at height Zr (m) (reference height for meteorological measurements), and Z0 is the “roughness length” at ground level, often assumed to be 0.01 m (0.4 in), representing short grass or a street.
     In addition to the environmental variables, UTCI also incorporates the metabolic rate of heat production and a detailed clothing model.  The clothing model is used to determine whole-body and local insulation values for the head, torso, lower arms, hands, upper and lower legs, and feet.  The vapor resistance of garments and air layers are also determined in the model.  Calculators available for practical application of UTCI do not require direct input of clothing characteristics; the model incorporates typical clothing appropriate for the environment.
     An online calculator is provided at utci.org with a simple interface.  If an offline option is desired, an executable file (“source code”) can also be downloaded from the site.  Repeated calculations in the program’s DOS interface can be tedious, however.
     For more information on the UTCI calculation process, there is a poster, also available on utci.org, that summarizes the operational procedure.  See Exhibit 5 for a preview of the UTCI summary poster.

     Like Heat Index and Wind Chill Index, an output of the UTCI model is a color-coded chart of thermal stress.  A simplified version appears in the visualization in Exhibit 4 and the summary poster in Exhibit 5.  The chart in Exhibit 6 provides additional details of typical physiological responses corresponding to various index temperatures within each stress category.  The “thermal comfort zone” is shown as the upper portion of the “no thermal stress” category, while there is no “slight heat stress” category defined.  The remaining categories – moderate, strong, very strong, and extreme – are mirrored for heat and cold stress.

     Returning to the de Freitas and Grigorieva scoring scheme, UTCI received a total score of 27, scoring the maximum for comprehensiveness, scope, sophistication, and transparency.  In this case, the scope score (first filtering criterion) was certainly deserved, with a cited valid range of -90 – 60° C (-130 – 140° F).
     The second filtering criterion, usability, scored only 3.  However, the rationale for increasing this to 5, acknowledging the availability of simple tools and calculators, once again applies.  This is the reason for its inclusion in the list despite its nominally deficient score.
     In the final category, validity, UTCI scored 4, reflecting the development team’s work comparing UTCI to several other indices.  This brings the revised total score to 29, placing UTCI alone atop the index-scoring hierarchy.  Scores alone, however, will not propel any index to a position of prominence in meteorological or industrial hygiene domains.  If it is to unseat the incumbent HI/Twc duo, there is much for UTCI’s advocates to accomplish.  Continued development of the index and its underlying models and assumptions are important.  Perhaps more difficult will be the education and persuasion of the general public and practitioners in several fields, cultures, and languages, whose motivations and capacities for change differ greatly.

     For additional guidance or assistance with management of thermal environments, or other Operations challenges, feel free to leave a comment, contact JayWink Solutions, or schedule an appointment.

     For a directory of “Thermal Work Environments” entries on “The Third Degree,” see Part 1:  An Introduction to Biometeorology and Job Design (17May2023).

References
[Link] “A comprehensive catalogue and classification of human thermal climate indices.”  C.R. de Freitas and E.A Grigorieva.  International Journal of Biometeorology; January 2015.
[Link] “A comparison and appraisal of a comprehensive range of human thermal climate indices.”  C.R. de Freitas and E.A Grigorieva.  International Journal of Biometeorology; March 2017.
[Link] “The Perceived Temperature:  The Method of the Deutscher Wetterdienst for the Assessment of Cold Stress and Heat Load for the Human Body.”  G. Jendritzky, et al. International Society of Biometeorology; 2000.
[Link] “A Universal Scale of Apparent Temperature.”  Robert G. Steadman.  Journal of Applied Meteorology and Climatology; December 1984.
[Link] “The Acclimatization Thermal Strain Index (ATSI): A preliminary study of the methodology applied to climatic conditions of the Russian Far East.”  C.R. de Freitas and E.A Grigorieva.  International Journal of Biometeorology; March 2009.
[Link] “New Indices to Assess Thermal Risks Outdoors.”  Krzystof Blazejczyk.  Environmental Ergonomics XI, Proeedings. of the 11th  International Conference; May 2005.
[Link] “An outdoor thermal comfort index (OUT-SET*) - Part I - The model and its assumptions.”  Richard de Dear and J. Pickup.  Proceedings of the 15th International Congress of Biometeorology and International Conference on Urban Climatology; January 1999.
[Link] "An Outdoor Thermal Comfort Index (OUT_SET*) - Part II – Applications."  Richard de Dear and J. Pickup.  Proceedings of the 15th International Congress of Biometeorology and International Conference on Urban Climatology; January 1999.
[Link] “Thermal Indices and Thermophysiological Modeling for Heat Stress.”  George Havenith and Dusan Fiala.  Comprehensive Physiology; January 2016.
[Link] “Threshold Limit Values for Chemical Substances and Physical Agents.”  American Conference of Governmental Industrial Hygienists (ACGIH); latest edition.
[Link] “UTCI - Universal Thermal Climate Index.”
[Link] “UTCI - why another thermal index?”  Gerd Jendritzky, Richard de Dear, and George Havenith.  International Journal of Biometeorology; December 21, 2011.
[Link] “The Universal Thermal Climate Index UTCI in operational use.”  Peter Bröde, Gerd Jendritzky, Dusan Fiala, and George Havenith.  Proceedings of Conference: Adapting to Change: New Thinking on Comfort Cumberland Lodge; April 2010.
[Link] “Deriving the operational procedure for the Universal Thermal Climate Index (UTCI).”  Peter Bröde, et al.   International Journal of Biometeorology; May 2012.
[Link] “The UTCI-clothing model.”  George Havenith, et al.  International Journal of Biometeorology; May 2012.
[Link] “The Universal Thermal Climate Index UTCI Compared to Ergonomics Standards for Assessing the Thermal Environment.”  Peter Bröde, et al.  Industrial Health; February 2013.
[Link] “Mean radiant temperature.”  Wikipedia.
 
Jody W. Phelps, MSc, PMP®, MBA
Principal Consultant
JayWink Solutions, LLC
[email protected]