272
]. A. ]. STOLWIJK
Table 3 that the controller coefficients which would represent additive
integration of different sensors are not used, whereas most of the coefficients
used represent integration of the multiplicative or gain control type.
By suitable adjustments of set point temperatures and controller coefficients, it is usually possible to obtain exact simulations of a particular
experiment. In fact, it is usually possible to do this with more than one
combination of set point temperatures and controller coefficients. The most
critical test of the predictive capabilities of a theoretical controller system
is the simulation of an experiment which is totally different in experimental
design and method from the ones on which the set point temperatures and
controller coefficients are based.
An example of such a test is shown in Figure 5 based on a series of
experiments performed by PIIRONEN [1]. The observed esophageal temperature is shown in steady state conditions in men working at different work
39.0
I
370
/ Iv°
------------ ---- --~O/!f{,x
-----_ __0
X I
0-->0.__ ~ ,
.-----7-. . . .
~x
...--x
x ____ x
38.0
25
27
29
31
33
35
37
1' . ee)
Fig. 5. Comparison of computer predictions of steady state rectal temperature
and ear temperature with measured esophageal temperatures in an experiment in
which subjects worked at 900 kpm at various average skin temperatures
Work level-900 Kpm, • Tr, x Te,
- - - Computer predictions, ------ Experimental results (2 subjects)
rates in environments which were so manipulated as to produce given average skin temperatures. Fig. 5 gives the computer prediction for rectal
temperature and head core temperature at a work level of 900 kgmjmin
for skin temperatures between 27°C and 36.5 0c. In the same figure are also
given the experimentally determined values of esophageal temperatures for
two of PIIRONEN'S subjects.
In the case of applied problems in body temperature regulation, predictions of this accuracy are quite useful. The model then presents itself as a