SPE 86606

Forecasting Energy Demand, Emissions and Discharges for the Petroleum Industry
Examples and Experiences
Kristin Keisers Bakkane, Novatech a.s, Geir Husdal, Novatech a.s, Marta S. Linde Melhus, the Norwegian Petroleum
Directorate, and Toril Re Utvik, Norsk Hydro

Copyright 2004, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at The Seventh SPE International Conference on
Health, Safety, and Environment in Oil and Gas Exploration and Production held in Calgary,
Alberta, Canada, 2931 March 2004.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
This paper presents state of the art methodology for
establishing reliable fuel consumption and emissions to air
forecasts for the offshore petroleum business in Norway. The
methodology is applicable for any operation within the
upstream oil and gas industry. It is unique due to a
combination of simple input, simple algorithms and accurate
output results.
The general forecasting method is established as a process
between the operators and Norwegian Authorities, under the
management by FUN (Forum for Forecasting and Uncertainty
Evaluations). The accuracy is proven by simple calibration
techniques, comparing measured fuel consumption against
calculated demands from using the forecasting method.
Forecasts have been established by Novatech on behalf of
the operator Norsk Hydro. Results for the Troll oil field are
shown as a sample case. The case verifies the ability to
forecast fuel consumption within an accuracy of 2-3% when
the forecasting model is checked by use of actual activity level
input data.
Also by the Norwegian Petroleum Directorate (NPD)'s
experience the precition of reported fuel and emission
forecasts has been gradually improved as the methodology as
described below has been implemented by the operators.
Background and Applications
Each year the operators on the Norwegian Continental Shelf
prepare fuel and emission forecasts to be reported to the
Norwegian authorities, as input to the Revised National
Budget (RNB). The data from the upstream oil and gas
industry are received and evaluated by the Norwegian
Petroleum Directorate (NPD).

Similar forecasts are required for a number of reasons, as

for instance as input to field development and operational
planning; company internal, field licensee or shareholder
reporting; emission permit applications and other
environmental concessionary requirements set by the
authorities, and for OPEX forecasting if any national taxes or
carbon trading systems apply.
All these purposes meet in a desire to make realistic
quantification of future fuel and energy demands, greenhouse
gas emissions etc. But for these forecasts to have any value,
they must be fairly accurate. For the oil and gas industry this
would as a minimum require a correlation to the activity
levels; i.e. basically the drilling level and the throughput of oil,
gas and water handled on an offshore installation or field.
Preferably it should also reflect any major production
philosophies and the technical design and constraints, or
perhaps rearrangements/replacements that may be foreseen
during the field production lifetime (or the forecasted period).
Ref. [1] describes a general methodology for doing all this,
and with means of relatively easily attainable information. The
method relates future emissions, fuel consumption and energy
needs to each other and to the oil, water and gas production,
injection and deliveries as well as the drilling activities
measured in number of wells drilled. Such forecasts are
normally readily available. Furthermore, the accuracy of the
fuel forecasts can be checked and the calculation models
involved may be calibrated to improve the accuracy if
required. Information from the forecasting model established
may even be used actively for operational optimization
purposes for a given case. This may actually identify potentials
for a reduction in fuel demand, reduced expenditure by lower
fuel consumption and possibly lower taxes, minimised
maintenance down-time, improved efficiencies, reduce losses
like for instance by flaring, etc.
The methodology is presented in the next chapter,
followed by a sample case and experiences gathered by the
second largest operating company on the Norwegian
Continental Shelf, Norsk Hydro, as well as by the Norwegian
Authorities, represented by the NPD who receives and
evaluates the annual RNB reporting from the operators.
General Forecasting Methodology
The general forecasting methodology presented in this paper is
based on the following simple concept:

SPE 86606

1.
2.
3.
4.

Forecast production and injection rates, etc.

Calculate energy demands based on these rates
Calculate fuel consumption based on this demand
Calculate emissions based on the fuel consumption

Any emissions and discharges supplementary to the emissions

from such energy consuming operations need to be added on
top of this to get the total emissions.
Energy Demanding Operations
An offshore installation involves a number of energy
demanding activities. By categorizing these by a typical
activity level describing unit, they can be divided into a
smaller list of operations, from here forward referred to as
Unit Operations. An example of such would be water injection
(the activity level described by the water throughput, which is
the quantity of water pumped for injection). The oil delivery,
gas injection and gas deliveries are three other examples.
Step 1 Forecast the Throughput (Production Rates etc.).
This part is normally taken care of by reservoar engineers
(water, oil and gas production rates), and/or in co-operation
with process engineers (gas and water injection, gas lift
forecasts). The well drilling and maintenance schedule will
also be an input on this first step.
The number one challenge during step 1 is to make sure
that the personnel involved in producing all of this input, make
their forecasts based on the same basis regarding major
technical and operational presumptions. This may for instance
involve phasing in production according to new wells,
removing gas reinjection rates when gas deliviery pipelines are
made available etc. Such presumptions should also be
communicated to personnel involved in the following steps:
Step 2 Energy Consumption (Energy Output). The energy
consumption over a time period for a unit operation can be
calculated from the throughput and an energy factor or
function. Mathematically, this may be expressed by the
following equation:
(1)
i
E
T
e

Ei = Ti ei
= unit operation indictor
= Energy consumption (the output as required for the
unit operation i)
= Throughput (production / injection volume etc.)
= energy factor or function

The energy factor or function represents a quantity of

energy required to handle a certain quantity of throughput for
the unit operation in question. Some examples would be:
T available in [m3], and E to be calculated as [MJ] or
[kWh] e to be established in [MJ/m3] or [kWh/m3],
respectively.
(Typical case: i = water injection)
T available in [number of wells drilled], and E to be
calculated as [MJ] or [kWh] e to be established in
[MJ/well] or [kWh/well], respectively.
(Typical case i = drilling)

The most challinging part of this stage is to establish the

energy factor or function. It is the first part of "the forecasting
model" for a unit operation.
ei can be established by using design data, if necessary
adjusting it to normal operation levels. It can also be found by
monitoring or calculating the actual (and total) kWh
consumption over a fixed period of time, for which the
throughput is known. The only restriction is to ensure that the
actual performance of the unit operation in the investigated
period is representative.
Step 3 Energy Demand (Energy Input). At this step one
needs to establish basic characteristics of the energy carrier
and the energy conversion machinery. Here the the lower
calorific value of the fuel and efficiency of the engine or the
generator/compressor/pump is taken into the equation. As a
result, the fuel requirement is calculated as follows:
(2)

1
1
Ci = Ei
and Fi, j = Ci
NCVj
i

C = energy input [MJ] or [kWh]

= the overall, average energy conversion efficiency
F
= fuel consumption [Sm3 gas fuel] or [tonne diesel]
j
= fuel type indicator (= fuel gas, diesel etc.)
NCV = fuel Net Calorific Value [MJ/Sm3 gas fuel] (or for
instance [kWh/tonne diesel] etc.)
In the efficiency factor shoul be representative for the
forecasting period in question, and all normal losses should be
taken into consideration. These are for instance gear losses,
generator losses, transfer losses etc.
The thermal efficiency should also correspond to the Net
calorific value of the fuel.
Step 4 Calculate the Emissions. This step is carried out by
multiplying the calculated fuel consumption forecasts by a
representative emission factor:
(3)
x
f
G

Gx = F f x, j
= CO2, NOX etc. (emission gas indicator)
= Emission factor [tonnes of x/Sm3 fuel gas] or
[tonnes of x/tonne of diesel]
= Emissions [tonnes of x]

The emission factor should correspond correspond to the

type of fuel (j). Beware that there might be cases where f also
depends on the unit operation i. There may also be emission
mitigation or cleaning technologies involved, and the factor
should reflect the effect of such technologies.
The Forecasting Model. Now the forecasting model for
unit operation i is complete, as follows:
(4)

1
1
Mi, j,x = ei

f x, j, i
NCVj
i

M = Calculation model for forecasting the emissions of x

based on fuel type j for unit operation i

SPE 86606

The model can also be illustrated as shown in Figure 1.

Fuel j
Fuel j NCV

Cleaning
Emissions
of x

Energy
consumption

e
Generator,
Generator,
pump etc.

pump etc.

measured for a representative period of operation and then

used as a basis for extrapolation.
Crude Oil Storage and Offshore Loading. With the
offloaded volumes of oil forecastes as for Step 1 above, the
challenge will be to establish reasonable emission factors. The
method should take necessary considerations as to the oil
vapour pressure, temperature conditions, the weather
conditions and more. This issue is rather extensive and will
not be handled here. Reference [3] is recommended.
A Sample Case

Figure 1 Relations between energy consumption and emission

The emissions Gijx (for the investigated combination of i, j

and x), are calculated as the product of this forecasting model
and the relevant throughput Ti.
Finally, the total emissions of any x (like CO2 etc.) from
energy consuming operations may be calculated as:

(6)

Case Descripion
The above described forecasting methodology has been used
for all fields operated by Norsk Hydro since 1998. One of the
fields is the Troll Field where Norsk Hydro is the operator of
the oil resources (Troll Olje), while Statoil operates the gas
resources by means of a gravity based production platform.
A layout of the field is shown in Figure 2.

Gx = Mi, j, x Ti, j ,
i

- where all relevant contributors (all fuel types j and all unit
operations i) must be included.
Other Emissions and Discharges
Other emissions and discharges would cover any source not
being generated due to an energy demanding operation. This
typically covers combustion processes like flaring and wellstream burning after well testing. Also unburnt hydrocarbon
emissions will apply here, like low-pressure venting and
fugitive emissions. For these emission sources the quantities
may be estimated as follows:
Gas Flaring. In this case, the combusted volume equals
the throughput. It has to be established by experience data,
based on any given operator philosophy or any other
reasonable estimate of annual quantities of gas flared. General
gas flaring emission factors are available from OLF (ref. [2]);
otherwise flare vendor data could be used, or the emissions
would have to be measured, and the emission factor
established thereof.
Oil and Gas Burning During Well Testing. The same
principle as for gas flaring can be used. General emission
factors recommended by OLF are found in reference [2].
Cold Vents and Fugitive Emissions. Again, the principle
is similar to gas flaring; only that the throughput will be the
total gas handeled on the installation in question. General
emission factors are established by identifying a number of
standardized sources for gas venting and fugitive emissions.
Once these sources have been established, they can be
combined with general emission factors as recommended by
OLF in reference [2]. Alternatively, the emissions can be

Figure 2 TheTroll field

The oil, which is present in a thin layer (1-15 m) between

the gas cap and the underlaying aquifer, is produced by a large
number of subsea completed wells. The wells are tied into two
semisubmersible production platforms, Troll B and Troll C,
where oil and associated gas are separated. The gas is dewpoint controlled on the platforms and sent by a gas pipeline to
the gas platform, Troll A, for further transportation to shore (at
Kollsnes). The oil is stabilized on Troll B and C and
transported by two pipelines to the Sture oil terminal.
The oil production is enhanced by gas lift, by gas injection
(Troll B) and by water injection at the Troll Pilot, which is a
seabed separation station. The other water production is
discharged to sea.
In the following the two installations on the field (Troll B
and Troll C) will be used as examples for description of:
Platform dedicated model
Emission forecast
Calibration of model
The results obtained

SPE 86606

The process of data collection and data flow in the

calculations
Model Description
The forecasting model for Troll Olje is split in two, one for
Troll B and one for Troll C. The operation taken place on both
platforms are almost identical, with some differences in
separation pressure and treatment volumes. The following unit
operations are used:
Oil treatment and delivery
Gas compression and delivery
Water injection (at Troll Pilot)
Gas lift
Gas injection
Well drilling (by mobile units only)
Platform base load
All wells are drilled by mobile drilling units, using diesel
oil for fuel. Gas compressors are powered directly by gas
turbines. All other operations are being powered by electric
power, supplied by the gas turbine driven generator stations.
Natural gas is used as fuel, except for platform cold start
operations where diesel fuel is used.
The energy correlations were established based on process
data provided by Norsk Hydros field operating group.
Relatively simple correlations between the throughput of each
unit operation and the energy demand were established. These
have been somewhat ajusted based upon feedback from the
operating organisation and the results that came out from the
annual calibrations.
The gas turbines are all GE LM 2500 standard air
derivative units. One of the generator turbines on each
platform is a dual fuel unit. The power generation efficiencies
for these units were established based upon expected future
turbine loads.
Fuel calorific values and emission factors were established
from fuel analyses and vendor data (for NOX and uncombusted
hydrocarbons).
The adjustments of gas turbine efficiency factors, fuel
energy data and emission profiles have been small ever since
the Troll models were established.

Figure 3 CO2 emission forecast for Troll B and Troll C, aggregated

and split on unit operations

Quite typical, this case shows that the main contributor to

the energy demand, and thus the CO2 emissions, is gas
compression.
Model Calibration
The model has to be calibrated to checked against errors and
improve the accuracy of the forecasts in coming years. Since
the emissions are not measured, just calculated, fuel gas
consumption is used as the calibrated parameter.
The calibration is performed on historical throughput data
(prodution, injection etc.), normally on a monthly basis. The
corresponding fuel gas consumption is calculated by use of the
established forecasting model, and then compared to the
measured data.
Minor deviations on a monthy basis are quite common and
illustrate that the model is a simplification of reality. There
will always be details that are not reflected in the model. The
question is whether such details are significant or not. If the
deviation is consistent, although small, it may be caused by a
minor error in the model input data. This could be the
correlation between energy demand and production or it can
be errors in fuel gas calorific value or in gas turbine efficiency
levels.
Figure 4 shows calibration data for Troll B. The curves
show minor deviations only (2-3% aggregated). Deviations of
this magnitude are normally not resulting in any specific
action.

Emission Forecasts
The fuel and emission forecasts have been calculated by
Novatech on behalf of Norsk Hydro based on throughput
forecasts from Norsk Hydro. A tailor made computer
programme which implements the above described
methodology is used to set up the forecasting models. The
CO2 emission forecast for the Troll Oil sample case is shown
in Figure 3, split on unit operations.

Figure 4 Calibration of fuel gas volumes for Troll B

SPE 86606

If deviations are significant, a more detailed analysis will

be required. Since the calculated fuel comsumption can be
split if the various unit operations, it may be relatively easy to
identify where the error(s) could be found. Sometimes,
however, deviations between calculated fuel consumption and
measured ones could be difficult to explain. It has been
obserbed that fuel consumption has gone up for one or for
some months, while none of the input profiles have shown
such behaviour. The reason may then be operational problems
etc. Normally there is always an explanation although it is not
obvious. This kind of deviations between the model and the
reality will always exist, and it is important to realize that the
model can and should not cope with all kind of irregularities.
The important message is that the model should be relatively
accurate when it comes to the longterm forecasts.
It is the fuel gas that is calibrated, not the emissions. The
correlations between emissions and fuel consumptions should
therefore be controlled as well. For CO2, this is relatively easy.
A given fuel gas has a given composition dictating the carbon
number of the fuel and hence the amount of CO2 emission per
unit of fuel used. Regular gas fuel analyses will therefore be
sufficient to establish the carbon number and the CO2
emissions. A problem will occur if the fuel gas composition
changes uncontrolled over time. This is a situation that could
badly affect the available accuracy in the fuel forecast since
the fuel energy content (the NCV) fluctuates correspondingly.
A well-known challenge relating to variable gas composition
applies for gas flaring as the composition of flare gas will vary
with source. A high acurracy in the emission forecasts from
gas flaring is therefore not to be expected.
Control of NOX emissions may be far more difficult. The
specific NOX emissions vary with gas turbine load and
conditions. NOX measurements in exhauts gas at relevant load
conditions should therefore be performed on a regular basis.
The vendor data are quite often not representative for the
actual on-site conditons.

Norsk Hydro's Experience and Use

Norsk Hydro is the operator on the Norvegian Continental
shelf for the following installations:
Oseberg Field Centre
Oseberg C
Oseberg D
Oseberg East
Oseberg South
Tune (subsea)
Brage
Troll B
Troll C
Njord
Heimdal and Vale (subsea)
Grane
Forecasts on emissions to air from these installations are
established annually for reporting to the Norwegian
governmental authorities for their Revised National Budget
(RNB). A computer based programme is used for the

calculations, implementing the forecasting methodology as

described in this paper.
Up to the mid-90's the calculations were based on more or
less inaccurate judgements. The standardized methodology
described above was implemented in 1998 for the first time
and has been continuously improved thereafter. The
improvements consisted of:
improved technical data on processes and equipment
improved forecasts for input (production profiles for oil,
gas and water)
improved variables used in calculations and calibration of
model. The deviations between calculated values and
historical values are in most cases less than 5%.
The data generated in the calculations are used for:
generating environmentally related input to the
Norwegian Revised National Budget (RNB)
environmental input to PDOs (Plan for Development and
Operation)
environmentally related reporting to the other licensees
forecasting quantities of potentially harmful chemical
substances discharged with produced water
The experiences from six years of emission forcasting, using
this methodology, are:
the emission profiles from plateau production through tail
end production were overestimated before this
methodology was implemented
forecasting emission profiles are today relatively accurate
the effects of phasing in new fields/finds are easily
implemented in the calculations
the annual updates can be performed effectively
The Forecasting Process
Norsk Hydro has used this methodology for all fields in
operation since 1998. An annual update is performed every
fall in connection with the reporting to the governmental
authority.
Norsk Hydro has experienced that emission forecasting is a
continuous process. It takes place in several phases and start
late winter when last years production and fuel data are
available. The steps in this process are described in Figure 5.
Initial Preparation of
Technical Input
Activity (Throughput)
Profiles

Planning the Forecasting

Process

Adjusting Technical Input

Parameters

Model
Calibration

Forecast
Calculation

Quality
Control

Figure 5 The fuel and emission forecasting process

SPE 86606

Once a model has been established, fuel calibration is the

first step in updating the forecasts. This is done by adding on
actual input data (measured throughput rates and fuel
consumption) for the period since last calibration was
performed. If the results from the calibration show significant
deviations between calculated and recorded fuel gas, the
model and technical input data are being analysed and
adjusted as necessary. At the same time the documentation is
updated. This documentation, which normally consists of a
text document and an associated spread-sheet, is the main tool
of comunication between the parties involved and shall
contain all required documentation of the forecasting model
(energy factors and all technical input data).
Calibration and documentation updating for all fields
operated by Norsk Hydro are performed over a period of 1-3
months. This work should normally be completed well in
advance of the forecasting work.
A very important exercise is to ensure that the forcasting
input data are established and put together in a consistent
manner to avoid errors and to simplify the required quality
control. This is specifically important for fields acting as hubs
with one or more satellite fields tied in for transport or final
processing. To ensure this a special matrix control sheet is
used. Quality control of input data is important to avoid errors
that may significantly affect the results.
The calculations can be performed on a spread-sheet or a
more dedicated computer program. Norsk Hydro has financed
development of such a dedicated programme. This software
has been used for all Hydros fields for the last three years.
After calculating the new forecast, quality control of the
output data is required. An important part of the quality
control is to compare the forecasts with the previous one and
explain the reasons for differences.
Major advantages of the methodology is that
it enables forecasting of energy demand, fuel demand
and emissions in one operation
it enables the energy demand, fuel demand and
emissons to be distributed on the source activities,
which is important for cost efficient mitigation
The whole process is illustrated in the Gannt diagramme
below, see Figure 6.
Activity
Technical input data preparation
Preparation of production profiles
Calculation of environmental data
Comments to the environmental data
Quality control
Comments from NPD/updates
Debriefing of process
Calibration of model and input data

Nov

Des

Proposed schedule
Jan Feb Mar Apr

Mai

Jun

Jul

Aug

Sep

Okt

Nov

Figure 6 Project planning in annual fuel and emission forecasting

Learning Points
The advantages of the calculations and use of the model are:
Forecasting of gas emissions, including greenhouse gases,
from offshore oil and gas production can be done quite
accurate. Consequently precise forecasts of GWP can be
made. The need for precise predictions is important
related to the Kyoto commitments and the operating

companies relative contributions to national emissions

CO2 taxes can be calculated based on today's tax value.
The taxes are based on the Sm3 of fuel gas used.
The software used in the calculation can also be used as a
what-if tool for analyzing different scenarios. This also
applies to simulations on quota trading
Data of good quality can be submitted to partners
The Government will have access to precise estimates on
emission forcasts from the offshore industry. This also
applies to estimates on national tax income from CO2
taxes from the offshore industry.
Norwegian Authorities' Experience
Each year the Norwegian Petroleum Directorate (NPD)
establish updated emission forecasts for the Norwegian
petroleum industry based upon data reported to the Revised
National Budget (RNB). The operators on the Norwegian
sector report the emission forecasts for each installation. NPD
calculates forecasts for the undiscovered recourses,
exploration activities, terminals and etc. Together this gives a
total emission forecast for the Norwegian petroleum sector.
The reporting process has been improved the last years.
The operator delivered reports on Secure Oil Information Link
(SOIL) network where the last updated version is available
both for NPD and the operator themselves. The quality of the
data has been improved because operators and authorities see
the importance of establishing emission data of high quality.
The authorities need a correct basis for making decisions
concerning national emission goals and international
engagements. A high quality emission forecast also makes it
easier for the operator to plan efforts and investments related
to emission permits.
One of the greatest challenges in the reporting process is to
retain the consistence between the emission data, the
production data and the economic data. Different emission
sources correlate to production data in different manner. For
instance, the main source for CO2 and NOx emissions is the
power generation. There is no linear correlation between the
power demand and the oil and gas production for an
installation. The knowledge of this has increased the quality
of the emission forecasts the last years.
NPD has focused on field specific factors to be used in the
calculations. The operators have prioritized this work and the
quality of the emission forecasts has then been improved.
Conclusions
The general forecasting methodology described in this paper is
capable of establishing accurate fuel forecasts based on key
activity level forecasts (production and injection forecasts etc)
and key technical information input. Once the forecasting
model is established, the forecasts are easily upgraded by
adjusting relevant input parameters. Furthermore, the
methodology fulfils the accuracy and transparency requirements for fuel and emission forecasting, of both operators and
Norwegian authorities.

SPE 86606

Nomenclature
OLF
= The Norwegian Oil Industry Association
NPD
= The Norwegian Petroleum Directorate
Sm3
= Standard cubic meters (at 15C and 1 bar)
FUN
= Forum for Forecasting and Uncertainty
Evaluations (working committee with
representatives from NPD and operating
companies on the Norwegian Continental Shelf)
RNB
= The Norwegian Revised National Budget
GWP
= The Global Warming Potential (normally
measured in CO2 equivalents)
i
= unit operation indictor
j
= fuel type indicator (= fuel gas, diesel etc.)
E
= Energy consumption (the output as required for
the unit operation i)
T
= Throughput (production / injection volume etc.)
e
= energy factor or function
C
= energy input [kWh] (input)
x
= CO2, NOX etc. (emission gas indicator)
f
= Emission factor [tonnes of x/Sm3]
G
= Emissions [tonnes of x]

= the overall, average energy conversion efficiency

F
= fuel consumption [Sm3 gas fuel] or [tonne diesel]
NCV
= fuel Net Calorific Value [kWh/Sm3 gas fuel] or
[kWh/tonne diesel] (lower heating value)

References
[1] FUN (Forum for Forecasting and Uncertainty Evaluations):
"FORECASTING GUIDELINES, Part I: Energy Consumption
and Emissions to Air in the Upstream Petroleum Industry.
General Methodology, Basic Input Data, Numerical Examples,
Planning of Work, Calibration of Forecasting Model", 2003
(may be downloaded from www.novatech.no)
[2] OLF "Veiledning for utslippsrapportering" ("Guidelines for
emission and discharge reporting"), 2001 (in Norwegian)
[3] Statoil, Norsk Hydro, EssoMobil and Norske Shell: "Joint
Industry Guidelines Determination and Verification of
NMVOC Emission Factors Offshore Crude Oil Storage and
Loading on the Norwegian Continental Shelf", 2001