Dependent Indicators for Environmental Evaluations of Desalination Plants
127
to exclude those indicators with incomplete data. This option is not acceptable for
the data presented in Table 3 because it will result in eliminating almost two thirds
of the indicators.
Another option is to exclude indicators that are not highly relevant to the
environmental objective. For instance, SiO 2 is considered by the World Health
Organization (WHO) as a safe chemical for marine environment, mainly because
it is a dietary requirement for various organisms. Hence, SiO 2 (indicator 14 in Table
3) can be safely excluded from the environmental analysis.
The data presented in Table 3 still has many missing values some adjustments are
needed before testing the assessment methodology presented above. The procedure
starts by studying the dependencies between indicators in order to complete missing
data or otherwise reduce the number of indicators. Examples of using dependencies
in estimating missing values for indicators are:
a) Measurements for the concentration of carbonate (CO 3
− ), which is indicator 15
in Table 3, is missing. But as indicated in Table 1, carbonate can be estimated
using the total alkalinity (At), i.e. [CO 3
−2 ] = 0.6 At. Furthermore, alkalinity
measurements are available for three plants (indicator 25 in Table 3).
b) Permanent water hardness (indicator 26 in Table 3) is expressed as equivalent
of CaCO 3 and this property is usually related to compounds with calcium and
magnesium ions (Ca ++ and Mg ++ ions). It is therefore possible to estimate the
missing total hardness measurement for plant (d) by applying the correlation in
Table 1, i.e. [CaCO 3 ] = 2.5[Ca 2+ ] + 4.1[Mg 2+ ].
c) Langelier Saturation Index (LSI) (indicator 22 in Table 3) is a calculated property
that reflects the stability of calcium carbonate in water. It estimates the saturation
level of calcium carbonate and indicates the extent of scale deposition on heat
transfer surfaces. Values of the LSI indicator is missing for the last three plants
in Table 3. The exact procedure for evaluating the LSI is relatively complex.
The alternative was to estimate the missing values of the LSI indicator using an
online LSI calculator reported by LENNTECH (2016). Required data include
acidity (pH), calcium ion (Ca ++ ), bicarbonate (HCO 3
− ), and total dissolved
solids (TDS), which are indicators 7, 1, 16 and 9 in Table 3, respectively, together
with brine temperature. Since the brine temperature is not reported, the LSI was
evaluated at 40 ◦ C, which is considered a typical brine temperature taken from
different studies (see for example Dawoud and Al Mulla 2012 and Kotb 2015).
To check the validity of the assessment steps presented above, the data provided
in Table 3 is divided into three datasets. The aim is to test the effect of indicator
dependencies, data sizes, and model format when ranking with the Copeland
method. The selected datasets are presented in Table 4. The first dataset includes
three objects (plants a, b and c) and 23 indicators, with no missing data. One of
the excluded indicators (SiO 2 ) is irrelevant, while the rest (Carbonate, LSI and
Hardness) are initially excluded from the analysis due to incomplete data. Applying
the Copeland ranking procedure on the (3 × 23) dataset that includes the three
plants a, b and c (Alssadanat, Umm Alquain and Hamriyah) results in normalized
ranks of 0.227, 1.0 and zero, respectively. The next step is to extend the original
127
to exclude those indicators with incomplete data. This option is not acceptable for
the data presented in Table 3 because it will result in eliminating almost two thirds
of the indicators.
Another option is to exclude indicators that are not highly relevant to the
environmental objective. For instance, SiO 2 is considered by the World Health
Organization (WHO) as a safe chemical for marine environment, mainly because
it is a dietary requirement for various organisms. Hence, SiO 2 (indicator 14 in Table
3) can be safely excluded from the environmental analysis.
The data presented in Table 3 still has many missing values some adjustments are
needed before testing the assessment methodology presented above. The procedure
starts by studying the dependencies between indicators in order to complete missing
data or otherwise reduce the number of indicators. Examples of using dependencies
in estimating missing values for indicators are:
a) Measurements for the concentration of carbonate (CO 3
− ), which is indicator 15
in Table 3, is missing. But as indicated in Table 1, carbonate can be estimated
using the total alkalinity (At), i.e. [CO 3
−2 ] = 0.6 At. Furthermore, alkalinity
measurements are available for three plants (indicator 25 in Table 3).
b) Permanent water hardness (indicator 26 in Table 3) is expressed as equivalent
of CaCO 3 and this property is usually related to compounds with calcium and
magnesium ions (Ca ++ and Mg ++ ions). It is therefore possible to estimate the
missing total hardness measurement for plant (d) by applying the correlation in
Table 1, i.e. [CaCO 3 ] = 2.5[Ca 2+ ] + 4.1[Mg 2+ ].
c) Langelier Saturation Index (LSI) (indicator 22 in Table 3) is a calculated property
that reflects the stability of calcium carbonate in water. It estimates the saturation
level of calcium carbonate and indicates the extent of scale deposition on heat
transfer surfaces. Values of the LSI indicator is missing for the last three plants
in Table 3. The exact procedure for evaluating the LSI is relatively complex.
The alternative was to estimate the missing values of the LSI indicator using an
online LSI calculator reported by LENNTECH (2016). Required data include
acidity (pH), calcium ion (Ca ++ ), bicarbonate (HCO 3
− ), and total dissolved
solids (TDS), which are indicators 7, 1, 16 and 9 in Table 3, respectively, together
with brine temperature. Since the brine temperature is not reported, the LSI was
evaluated at 40 ◦ C, which is considered a typical brine temperature taken from
different studies (see for example Dawoud and Al Mulla 2012 and Kotb 2015).
To check the validity of the assessment steps presented above, the data provided
in Table 3 is divided into three datasets. The aim is to test the effect of indicator
dependencies, data sizes, and model format when ranking with the Copeland
method. The selected datasets are presented in Table 4. The first dataset includes
three objects (plants a, b and c) and 23 indicators, with no missing data. One of
the excluded indicators (SiO 2 ) is irrelevant, while the rest (Carbonate, LSI and
Hardness) are initially excluded from the analysis due to incomplete data. Applying
the Copeland ranking procedure on the (3 × 23) dataset that includes the three
plants a, b and c (Alssadanat, Umm Alquain and Hamriyah) results in normalized
ranks of 0.227, 1.0 and zero, respectively. The next step is to extend the original