You are on page 1of 14

UNIVERSITY OF MANCHESTER

School of Mechanical, Aerospace & Civil Engineering

1st Year Thermodynamics Laboratories on Working Fluids

Report on Laboratory 1: Principles of Saturation Pressure


Measurement for Steam and Analysis of Experimental Errors

Prepared by

Shuhaib Maudarbaccus
st
1 Year Mechanical Engineering

Experiment performed on: March 19, 2010


Report submitted on: April 19, 2010
Table of Contents
Abstract ..................................................................................................................................... 2
Introduction .............................................................................................................................. 3
Method ...................................................................................................................................... 4
Apparatus ............................................................................................................................... 4
Procedure ................................................................................................................................ 5
Results and Discussions ........................................................................................................... 6
Comments on Graphs of Absolute Temperature against Absolute Pressure ......................... 8
Comments on Heating and Cooling Rates ............................................................................. 9
Calculation of Experimental Error in the Reading of Barometric Pressure ......................... 10
Conclusions ............................................................................................................................. 11
Appendix I – Background Theory for the Experiment ...................................................... 12
Appendix II – Estimation of the Error in a Measurement................................................. 13
References ............................................................................................................................... 13

List of Figures
Figure 1: Labelled diagram showing apparatus setup................................................................ 4
Figure 2: Graph of Tabs and Tact against Pabs for Heating Stage ................................................. 7
Figure 3: Graph of Tabs and Tact against Pabs for Cooling Stage ................................................. 7
Figure 4: Graph of measured absolute temperature against elapsed time ................................. 9
Figure 5: Graph of measured absolute temperature against elapsed time, including lines of
best fit......................................................................................................................................... 9

Figure 6: p-T graph for water at constant volume ................................................................... 12


Figure 7: T-p graph for water at constant volume ................................................................... 12

List of Tables
Table 1: Experimental data with absolute temperature and pressure readings .......................... 6
Table 2: Data for heating stage .................................................................................................. 6
Table 3: Data for cooling stage .................................................................................................. 6
Table 4: Data for absolute temperatures and elapsed time ........................................................ 8

1
Abstract
This experiment was carried out to investigate the effect of response time of sensors or
thermal lag due to apparatus on the accuracy of the measurements of a property. The property
in this case was the saturation temperature of water measured under different values of
pressure. It was believed that the difference between heating and cooling rates would reveal
the effect of thermal lag or response time but this was proved untrue from analysis of the
results. It has been concluded, instead, that the accuracy of the results was mostly affected by
a systematic error which seemed to have been introduced by the way in which the readings
were recorded. However, only further experimentation, with some improvements, can verify
this conclusion.

2
Introduction
The purpose of the experiment was to investigate the effect of response time or thermal lag
on the accuracy of measurements. This meant that the results would need to be compared
with some theoretical values. The use of a property allowed this comparison to be made
provided other quantities were known. In this case, the saturation temperature of water was
investigated at different pressures but at constant volume. From the two-property rule, since
pressure readings could be obtained from the pressure sensor and volume was maintained
constant (a closed boiler was used), the third property which is the saturation temperature
could easily be determined from the tables for the thermodynamics properties of water.

Hence, with both experimental and theoretical values determined, the results were analysed
through the use of graphs. However, to be able to investigate thermal lag or response time,
the experiment had to be carried in two or more different conditions so that the results could
be compared. This was achieved by considering the variation of saturation temperature
during heating and cooling, as heating and cooling rates should normally be different.

The graphs for the heating and cooling stages were then analysed so that conclusions on the
effect of thermal lag or response time could be reached. The idea that heating and cooling
rates are different was also studied during the analysis.

3
Method1
Apparatus

Figure 1: Labelled diagram showing apparatus setup

The apparatus consists of a boiler connected to a network of pipes and valves, as shown in
Figure 1, which allows the boiler to be filled with the test substance, the air in it to be pushed
out by the vaporised substance, or the pressure inside it to be relieved if need be – for
example, to quickly reduce the pressure in case of emergency.

A platinum resistance thermometer is used to detect changes in temperature through an


electrical output in Ohms. Such a thermometer has a wide temperature range and gives
excellent accuracy. It can operate under a wider range of pressures than other types of
thermometers but requires a protective shield in liquid and corrosive atmospheres. Such
shielding does increase the response time of the sensor and the size of the sensor probe.

An electronic pressure sensor of semiconductor type has been used. In this sensor, one side of
a diaphragm is exposed to the pressure to be measured, while the other side is open to
atmosphere. The resulting deflection forces a rod into a metallic strip with semiconductor
resistance gauges bonded to the surface. The resulting tension or compression in these gauges
produces a measurable change in the semiconductor resistance, which can then be converted
to a pressure reading by a suitable conditioning circuit.

4
A conventional Bourdon-type gauge has been included to give a visual indication of the
pressure inside the equipment. This is intended as an extra safety measure, indicating when
the system is pressurised even if the water is not visibly active. Bourdon gauges can be
constructed to cover a wide pressure range but due to the nature of the display, the changes
will affect the accuracy of the sensor – in cases where temperature variation is pressure-
dependent, the sensor may be calibrated to compensate. The gauge is relatively bulky and
robust, and output purely mechanical. The accuracy of the measurements taken will be
affected by the properties of the sensors chosen.

The console used in the experiment provides a variable heater control and a three-way switch
controlling the digital readout with respect to the quantity being measured. In this
experiment, only the PT100 (1) and PRESSURE switch positions need to be used. The digital
readout gives resistance readings to the nearest 0.1 Ω and pressure readings to the nearest
1kPa. However, note that the meter has not been properly calibrated and that correction and
reference charts need to be used with the resistance readings while zero error needs to be
considered in the pressure readings.

Procedure

Once the setup of the apparatus had been checked, the console was switched on. Using the
PT100 (1) and PRESSURE switch positions, the initial resistance and pressure readings could
be found. The measurement of the initial pressure revealed a zero error of -16 kPa.

The heater was then switched on to the maximum heating level and a stopwatch was started
at the same time. At intervals of 2 minutes, the thermometer output and the reading from the
electronic pressure sensor were recorded. Since the safe pressure limit of the apparatus is 7
bars, the last set of readings to be recorded was when the Bourdon gauge indicated a pressure
just below 7 bars.

The heater was then switched off and the same set of readings taken at intervals of 5 minutes.
Due to lack of time, only 4 sets of values were recorded for the cooling stage.

The measured outputs from the platinum resistor were corrected using the Resistance Bridge
Correction Chart and the respective values of temperature determined using the PT100
Platinum Resistance Thermometer Reference Chart. Interpolation was used to improve
accuracy of the corrected values. The absolute pressure readings were determined by adding
16 kPa, due to the zero error, and the barometric pressure of 100.8 kPa to the readings from
the electronic pressure sensor.

5
Results & Discussion

Table 1: Experimental data with absolute temperature and pressure readings

Elapsed Measured Corrected Absolute Absolute Tabulated


Pressure
Time Output Output Temperature Pressure Temperature
t Rml Rcl Tabs P1 Pabs Tact
2 2
(mins) (Ω) (Ω) (K) (kN/m ) (kN/m ) (K)
0 138.8 138.47 373.15 -10 106.8 374.59
2 142.9 143.42 388.15 50 166.8 387.74
4 145.8 147.05 395.15 113 229.8 397.82
Heating

6 148.7 150.78 405.15 191 307.8 407.49


8 151.3 154.23 415.15 281 397.8 416.54
10 153.6 157.36 423.15 383 499.8 424.93
12 155.7 160.29 431.15 500 616.8 433.10
14 157.7 163.13 439.15 630 746.8 440.68
0 157.7 163.13 439.15 630 746.8 440.68
6 154.3 158.33 425.15 412 528.8 427.08
Cooling

11 151.1 153.96 414.15 267 383.8 415.23


16 148.3 150.13 403.15 167 283.8 404.81
20 146.0 147.30 397.15 111 227.8 397.54

Table 1 shows the recorded experimental data together with the corrected output readings
from the resistor (Rcl), the absolute temperature (Tabs), the absolute pressure (Pabs), and the
actual absolute temperature (Tact), determined from the “Thermodynamic and Transport
Properties of Fluids” handbook and considering the known values of absolute pressure.

The readings for the heating and cooling stages have been separated as shown in Table 2 and
Table 3 respectively. Percentage errors in the measured saturation temperature with respect to
the actual saturation temperature have also been included in the tables.

Table 2: Data for heating stage Table 3: Data for cooling stage
Pabs Tabs Tact Pabs Tabs Tact
% Error % Error
2
(kN/m ) (K) (K) (kN/m ) 2
(K) (K)
106.8 373.15 374.59 0.38 746.8 439.15 440.68 0.35
166.8 388.15 387.74 0.11 528.8 425.15 427.08 0.45
229.8 395.15 397.82 0.67 383.8 414.15 415.23 0.26
307.8 405.15 407.49 0.57 283.8 403.15 404.81 0.41
397.8 415.15 416.54 0.33 227.8 397.15 397.54 0.10
499.8 423.15 424.93 0.42
Average 0.31
616.8 431.15 433.10 0.45
746.8 439.15 440.68 0.35
Average 0.41
6
450

440

430
Absolute Temperature (K)

420

410
T-abs
400 T-act

390

380

370
0 100 200 300 400 500 600 700 800
Pabs (kN/m2)

Figure 2: Graph of Tabs and Tact against Pabs for Heating Stage

445

440

435
Absolute Temperature (K)

430

425

420
T-abs
415
T-act
410

405

400

395
200 300 400 500 600 700 800

Pabs (kN/m2)

Figure 3: Graph of Tabs and Tact against Pabs for Cooling Stage
7
Comments on Graphs of Absolute Temperature against Absolute Pressure

Both graphs in Figures 2 and 3 show a slope with a gradient which tends to decrease as
absolute pressure increases. This shape is the correct one since:
(i) it matches closely with the shape of the graphs of actual saturation temperature against
absolute pressure (shown dashed), and
(ii) absolute temperature (Tabs) was plotted against absolute pressure (Pabs) - this should
result in the reflection of the graph of Pabs against Tabs in the line Pabs = Tabs as shown in
Appendix I.
Note that this shape is less apparent in Figure 3 as only 5 sets of readings could be recorded
within the time available.

The graph in Figure 2, which is that of the heating stage, almost overlaps that of Tact (shown
dashed) between the 2 initial readings (at Pabs = 106.8 kPa and Pabs = 166.8 kPa). This can
accounted for by the fact that when the experiment was started, the apparatus together with
the sensors were in thermal equilibrium and therefore the initial readings obtained were not
affected by response time of the sensors or thermal lag due to apparatus. However, as further
readings are recorded, the two curves part off and the graph of Tabs remains below that of Tact
which could be due to high response time of sensors or thermal lag. Also, it is worth noting
that the third reading (at Pabs = 229.8 kPa), the temperature reading suddenly goes below the
actual temperature.
Analysing Figure 3, the graph for the cooling stage of the investigation, it is seen that once
again the graph of Tabs lies below that of Tact, which is not consistent with the previous
conclusion that the difference is due to high response time or thermal lag. Had it been the
case, the graph of Tabs should have been above that of Tact for the cooling stage. It can also be
seen that the last point recorded (at Pabs = 227.8 kPa) almost lies on the graph of Tact.

The above analysis therefore leads to the idea that there Table 4: Data for absolute
could have been a systematic error while doing the temperatures and elapsed time
experiment as the experimental values seem to differ Absolute Temperature(K)
from the theoretical values in a similar way for both Elapsed
Time Heating Cooling
heating and cooling stages. This error could have been (mins) Stage Stage
made when taking the readings from the console or
could also be a wrong calibration of the meters and the 0 373.15
charts used. Furthermore, it could be that some air was 2 388.15
present in the boiler throughout the experiment. 4 395.15
However, this does not seem coherent with the results 6 405.15
as it should have led to Tabs being higher and not lower. 8 415.15
10 423.15
Further investigation could be made by analysing the 12 431.15
graphs of measured absolute temperature against time. 14 439.15 439.15
The analysis is shown on the next page. 20 425.15
25 414.15
30 403.15
Table 4 shows the absolute temperatures (K) and the
34 397.15
elapsed time (minutes) when they were recorded.
8
Comments on Heating and Cooling Rates
450
Absolute Temperature (K)

440 A
430
420 Absolute
Temperature
410 Heating Stage
400
Absolute
390 Temperature
380 Cooling Stage
370
0 10 20 30 40
Elapsed Time (minutes)

Figure 4: Graph of measured absolute temperature against elapsed time

The graph in Figure 4 has been generated using data in Table 4. The graph reveals that the
temperature variation during heating and cooling is almost linear at high temperatures
(>410K). While the variation during heating at lower temperatures can be approximated as
linear (considering the randomness of the changes), the rate of cooling below 410 K seems to
decrease slightly as indicated by the curved part in the encircled region A. This could be the
result of a decreasing temperature gradient between the apparatus and its surroundings, hence
slowing down the cooling rate. However, this idea still needs to be confirmed by recording
more sets of values at lower temperatures to get a better understanding of the variation.

450
Absolute Temperature, y

440 Absolute Temperature


y = -2.129x + 468.12
430 Heating Stage
420
Absolute Temperature
(K)

410 Cooling Stage


400
390 Line of best fit
y = 4.5893x + 376.65 (Heating Stage)
380
370 Line of best fit
0 10 20 30 40 (Cooling Stage)
Elapsed Time, x (minutes)
Figure 5: Graph of measured absolute temperature against
elapsed time, including lines of best fit

Figure 5 shows the same graph as in Figure 4 but also includes the lines of best fit drawn for
both the heating and cooling stage, approximated as linear for the sake of comparison only.
By comparing the gradients of the best fit lines (gradient for heating stage = 4.589 and
gradient for cooling stage = -2.129), it can be deduced that heating rate is more than 2 times
faster than the cooling rate. In the light of this deduction, further conclusions may be reached
concerning the differences between the measured saturation temperature and the actual
saturation temperature.

9
Calculation of Experimental Error in the Reading of Barometric Pressure

Based on the example given in Appendix II, the error in the measurement of barometric
pressure can be estimated as follows:

The pressure, p, is given by the equation ‫ߩ = ݌‬. ݃. ℎ and the only measurement required is
that of the height, h, of the mercury column in the barometer, assuming that the density, ρ,
and the acceleration due to gravity, g, are both constants whose values are already
known to be 13.55 g/cm3 and 9.81 m/s2 respectively.

Also assume that the best accuracy that can be obtained from the built-in rule of the
barometer is to ½ of one graduation, which is equivalent to 5/1000 cm. The height, h, was
then found to be equal to 75.380 cm.

10଺ 75.830
‫ = ݌‬13.55 ∗ ଷ
∗ 9.81 ∗ = 100797.407 ܲܽ
10 100

5
‫ ܘ܌‬dρ dg dh 0 0
= + + = + + 1000 = ૟. ૞ૢ૝ ∗ ૚૙ି૞ = ૟. ૞ૢ૝ ∗ ૚૙ି૜ %
‫ܘ‬ ρ g h 13.55 9.81 75.830

∴ ࢊ࢖ = ૟. ૞ૢ૝ ∗ ૚૙ି૞ ∗ ૚૙૙ૠૢૠ. ૝૙ૠ = ૟. ૟૝ૠ ࡼࢇ

The error in pressure, p, is therefore estimated to be ± 6.647 Pa, which is not very significant
relative to the value of pressure itself (p = 100797.407 Pa).

It can therefore be deduced that the systematic error in the results did not arise as a result of
an error in the barometric pressure, the measurement being very accurate as seen above.

10
Conclusions
Having analysed the results of the experiment, it can be said that the hypothesis which relates
the difference between measured saturation temperature, Tabs, and actual saturation
temperature, Tact, to the effect of response time of sensors or thermal lag does not apply in the
cooling stage as seen in Figure 3. This implies that there should be a systematic error which is
affecting the results since the latter differ in the same way for both the heating and cooling
stages (graph of Tabs lies below that of Tact in both cases – refer to Figures 2 and 3).

A possible systematic error could have been introduced by the way in which the thermometer
output and reading from the pressure sensor were recorded every time. In fact, since the
switch positions need to be changed to get the required readings, this means that there is a
delay in reading off both quantities and the larger the delay, the bigger the difference between
Tabs and Tact. Besides, the order in which the results were being recorded could also affect this
difference. For instance, if in a particular case, the pressure was read first followed by the
resistance but in the next case, the resistance was read first and the pressure last, this would
lead to a variation in the difference obtained and this could account for the sudden deviation
of the curves at the third point in Figure 2.

Another systematic error could simply be the result of a wrong calibration of the console or
wrong use of the calibration charts in a systematic way (e.g. wrong method used for
interpolation). This would also be a very good explanation for the difference between
theoretical and experimental results.

Analysis of the heating and cooling rates (Figures 5) clearly indicated that heating rate was
much faster than cooling rate. Comparing the percentage errors for heating and cooling
stages, it was found, from Tables 2 and 3, that the heating stage has an average of 0.41%
while the cooling stage has one of 0.31%. This again brings up the idea that thermal lag and
response time could have been less apparent in the cooling stage, hence the slight difference
in average percentage errors. However, it could also mean that the effect of the delay due to
the first systematic error explained could have been reduced with the lower cooling rate. On
the other hand, the second systematic error explained cannot account for this difference.

It can therefore be concluded that the difference between the results were mostly affected by
a systematic error introduced when taking the readings and that the effect of thermal lag or
response time of sensors accounted for a minor part of this difference.

This conclusion can be verified by the use of data logging to record both readings
simultaneously, by repeating the experiment at lower heating rates to further investigate
effect of thermal lag or response time and by taking more readings during the cooling stage to
obtain a better estimation of the percentage error.

11
Appendix I – Background Theory for the Experiment2
The properties of water at constant volume can be represented as a function of pressure and
temperature as shown in Figure 6 below (alternatively, this curve can be seen by looking in
the direction of the p-T axes on the p-v-T surface block model found in the laboratory). The
saturation point of water is the condition at which a phase change occurs from liquid to
vapour, or vapour to liquid. It occurs at a very precise set of conditions, which form a line
when plotted on a graph such as that shown below (see the region marked L V).

Figure 6: p-T graph for water at constant volume

In this experiment, pressure is the


independent variable and temperature is the
dependent variable and therefore, a graph of
Temperature against Pressure needs to be
drawn.

Figure 7 shows the p-T graph above which


has been reflected in the line p=T to show
the T-p variation. The region being
investigated remains the same and should
lie between points X and Y encircled on
Figure 7. This gives an indication of the
shape of the graph to be expected from the
experiment and can help to confirm the
results.

Figure 7: T-p graph for water at constant volume

12
Appendix II – Estimation of the Error in a Measurement3
Consider the case where a rule graduated in millimetres is being used to read the dimensions
of a body nominally of length 300mm, width 100mm, and height 50mm in order to find its
volume. If it is assumed that the accuracy that can be obtained is to ½ of a graduation in the
measurements on the rule (this is a standard approximation), then the error in the
measurement of the volume is determined as follows:

ܸ =݈∗‫∗ݓ‬ℎ

where V is the volume. Take natural logs of both sides and differentiate to obtain

݈݊(ܸ) = ݈݊(݈) + ݈݊(‫ )ݓ‬+ ݈݊(ℎ)


ܸ݀ ݈݀ ݀‫݀ ݓ‬ℎ
= + +
ܸ ݈ ‫ݓ‬ ℎ

dV/V is the fractional error in the volume measurement and it can be seen that it is related to
the fractional errors in length, width and height measurement. Furthermore, each of these
errors is additive. Putting in the numbers (½ graduation is equivalent to ½ mm), it is found
that:

1 1 1
ܸ݀ 1 1 1
= 2 + 2 + 2 = + + = 1.67 %
ܸ 300 100 50 600 200 100

Notice that the worst error comes from the smallest value taken (i.e. the height of 50 mm in
this case).

References
[1] [2] [3] University of Manchester, School of MACE. [anon]. 1st Year Thermodynamics
Laboratories on Working Fluids, Laboratory 1. 2010

13

You might also like