This document discusses optimizing geothermal systems through dynamic systems modeling. It notes that key factors like well location, depth, flow rates, and reservoir properties need to be determined to assess power outputs. A dynamic modeling approach is presented that allows probabilistic simulation of how varying parameters like flow rates and drilling depth impact electrical power production over time, as higher flow rates can cause earlier temperature drawdown reducing long-term outputs. The modeling can help optimize design parameters and assess scenarios to maximize power for a given investment level.
3. www.preene.com
PRACTICE PROFILE
Preene Groundwater Consulting is the Professional Practice
of Dr Martin Preene and provides specialist advice and design
services in the fields of dewatering, groundwater engineering
and hydrogeology to clients worldwide
Dr Martin Preene has more than 25 years’ experience on
projects worldwide in the investigation, design, installation
and operation of groundwater control and dewatering
systems. He is widely published on dewatering and
groundwater control and is the author of the UK industry
guidance on dewatering (CIRIA Report C515 Groundwater
Control Design and Practice) as well as a dewatering text book
(Groundwater Lowering in Construction: A Practical Guide to
Dewatering)
4. www.preene.com
INTRODUCTION
• Many technical factors affect the development of geothermal
systems
• These are important but may only be indirectly related to the
project objective of maximising power (electricity and heat)
generation while minimising cost per unit power
• Quantity of power that can be generated over the project
lifetime is also important
• Parasitic losses can be important
• The whole system must be assessed and, if possible,
optimised
5. www.preene.com
WHY OPTIMISE?
• An understanding of optimisation is important at project
development stage to aid the development of a scheme that
maximises net power output for a given level of investment
• System optimisation is also important when looking at
forward predictions of net saleable power during funding
transactions or when agreeing power purchase agreements
• Several cost optimisation models exist
• There are some drawbacks and limitations with cost
optimisation models. Our approach is to focus on optimising
power outputs, to provide information to be used in financial
models
7. www.preene.com
KEY FACTORS
• Once a geological prospect has been identified, various key factors
must be determined before potential power outputs can be
assessed:
• Location (where to drill), and distance between extraction and
re-injection wells
• Depth of drilling
• Power conversion technology
• Mass flow rate (pumping and re-injection rate)
• Parasitic losses
• Reservoir pressure drawdown
• Reservoir temperature drawdown
8. www.preene.com
KEY FACTORS
• Parasitic losses
- Generating and cooling system parasitic loads
- Artificial lift parasitic loads
- Others
• Pressure drawdown
- Significant reduction in geofluid pressure will occur at extraction wells; this strongly
influences pumping parasitic losses
- Impact of pressure drawdown can be expressed as well productivity index =
production rate/drawdown
- Productivity index will be lower at higher mass flow rates, and may reduce with
time
• Temperature drawdown
- Geofluid circulation through the reservoir may reduce reservoir temperature in the
long term
- High mass flow rates may cause more rapid temperature drawdown and reduce
cumulative power production over defined periods
9. www.preene.com
DYNAMIC SYSTEMS MODELLING
• Most simple, and many relatively complex, systems can be
handled by spreadsheet based analysis, but it can be
difficult to capture options, uncertainty and interactions
• Tools like GoldSim are modelling environments for
probabilistic (Monte Carlo) simulation of complex dynamic
systems. These models are able to interact with other
modelling environments to produce coupled models
• In ‘Player’ mode, GoldSim can act as an interface for ‘non
technical’ end users to investigate change in key system
parameters
12. www.preene.com
EXAMPLE MODELLING
• Model outputs for a system based on binary power
conversion, with a single doublet of extraction and re-
injection well
• Key external parameters are geothermal gradient and
reservoir hydraulic properties (can be assigned a probability
density function)
• Key ‘optimisable’ parameters are depth of drilling and
volumetric flow rate (can be varied within a defined range)
• Model can be used to look at time series relationships and
parameter relationships
13. www.preene.com
MODELLING
• Permeable sandstone aquifer (lower end of hydrofractured systems
in terms of permeability).
• Well depth of 4.5 km.
• Geothermal Gradient of c. 0.047 C/m.
• Well spacing of 200 m.
• Mass flow rates between 10 kg/s and 50 kg/s.
• Thermal ‘cut off’ at 120 C (not reached).
• Run for 450 iterations.
• Rest water level 1380 m below ground level.
• Binary plant rejection temperature 330 K (57C).
• District heating circuit (final) rejection temperature 290 K (17C).
• Cooling load taken as 5% of gross electrical power output
15. www.preene.com
TIME SERIES MODELLING
ailable at plant (MWth)
Median (50%ile)
25th to 75th %ile
5th to 25th and 75th to 95th %ile
<5 %ile, >95 %ile
Power decreases as
reservoir cooling occurs
19. www.preene.com
OPTIMISATION OF PARAMETERS
•
Simulation realisations
At later times, temperature
drawdown has reduced geofluid
temperature (and therefore
power production) at high flow
rates
Net Electrical Power versus Abstraction after 300 Days
0
1000
2000
3000
4000
5000
6000
0 10 20 30 40 50 60
Mass Flow Rate (kg/s)
NetPower(kW)
Net Electrical Power versus Abstraction after 3000 Days
0
500
1000
1500
2000
2500
3000
3500
0 10 20 30 40 50 60
Mass Flow Rate (kg/s)
NetPower(kW)
20. www.preene.com
OPTIMISATION OF PARAMETERS
•
Simulation realisations
At later times, temperature drawdown has
reduced geofluid temperature (and therefore
power production) at high flow rates
Net Electrical Power versus Abstraction after 3000 Days
0
500
1000
1500
2000
2500
3000
3500
0 10 20 30 40 50 60
Mass Flow Rate (kg/s)
NetPower(kW)
21. www.preene.com
OPTIMISATION OF PARAMETERS
Simulation realisations
At higher flow rates,
temperature drawdown of
geofluid occurs earlier. The
temperature drawdown
reduces gross thermal
power and reduces
conversion efficiencies
22. www.preene.com
CONCLUSION
• Prediction of saleable power from geothermal systems involves a
complex series of interactions
• Involves uncertainty in external factors (e.g. geothermal gradient
and reservoir properties)
• Involves selection of controllable parameters (e.g. well depth, mass
flow rate) to optimise desired targets
• A dynamic systems approach allows predictive modelling of
potential resource and utilisation
• Can be used for scenario assessment during feasibility, funding or
project development stages. Can feed directly into financial models
We have heard in the earlier presentations about some of permitting factors and technical issues that need to be addressed in the development of a DGS.These are all important, but are only indirectly related to the driver for any project - to generate electricity and/or heat in a saleable form, in sufficient quantities and at low enough cost to make the project economically viable.It is not just about unit cost of saleable power, quantity over the project lifetime is important, because the obligations of Power Purchase Agreements must be metParasitic losses, for example running well pumps, cooling systems, will reduce net saleable powerThere is a need to assess the whole system and optimise where possible
It is stating the obvious to say that geothermal systems are complex. This diagram is relevant to a mid enthalpy system with a binary power conversion system and artificial lift pumps in the well. In reality this is one of the types of system that has the widest potential application across Europe.The primary elements of the system are easy to identify – the geothermal reservoir, the wellfield, the power conversion system (the turbine).But there are secondary elements which can also be important – artificial lift pumping systems in the wells, cooling systems feeding the turbine condensers, waste heat systemsIt does not make sense to look at these in isolation because they interact.
Key factors Need to drill in the right location, that’s why reconnaissance and feasibility studies are carried outIn general the deeper you drill the higher the bottom hole temperature, and the greater the power potentialGeofluid temperature has a big impact on power conversion. Is it hot enough to generate electricity? Is a binary plant necessary? Is it hot enough to flash to steam and be used directly in the turbine? In general power conversion efficiency increases with increasing temperaturePotential thermal output from a well increased with mass flow rate (kg/sec) from the well. The harder the pump, in theory more power is potentially available.It sound like it is as simple as drilling as deep as possible (max temp) and pumping as hard as possible (max thermal output). However it is not that simple.
Mention that GoldSim was originally developed by Golder AssociatesMuch more flexible than traditional linked spreadsheets
This is a GoldSim model set up for an example system comprising a doublet of one extraction well and one re-injection well, with electricity generation from a binary plant, and waste heat made available to a district heating system.You can see that it looks very similar to the system diagram I showed earlier. The flows and interaction are clearly traceable. In addition to a modelling tool it becomes a communication tool to share things within a design team.Each container contains a set of relationships, which may be very simple or complex. As the project evolves the contents of a conatiner can be changed. E.g moving from an analytical reservoir model to a numerical model.
Drilling depth, in combination with geothermal gradient, controls bottom hole temperature, which as a big impact on energy conversion efficiencyVolumetric flow rate (which is corrected by variable fluid density to mass flow rate) controls gross thermal power and has a big influence on pressure and temperature drawdown at the wellsNow we will look at some typical results
Example of GoldSim model output. Blue line is median (most likely) outcome, other lines are percentile predictionsBased on 4.5km deep well, with bottom hole temperature of 220 degrees C. Mass flow rate range from 10 to 50 kg/s (median 30 kg/s). Range of aquifer permeability values used. 100% of geofluid is re-injectedInitial geofluid temperature is around 220 C. Median line (30 kg/s) shows temperature drawdown is apparent after around 2,500 days, and after 10 years the abstracted geofluid temperature has fallen by around 30 C. For 50 kg/s temperature drawdown may after 1,500 days, and temp falls by around 70 C after 10 years. 10 kg/s shows no temperature drawdown after 10 years
Gross thermal power based on input and output temperatures at turbine30 kg/s shows around 30 MW. 50 Kg/s and favourable aquifer conditions show around 40 MW. But higher mass flow rates result in declining temperature and available thermal outputs
This is before parasitic losses, and takes account of cycle efficiency of turbine. Gross electrical power much lower than gross thermal power.Median power output around 4.5 MWCombinations of higher flow rates/and or unfavourable aquifer parameters show declining power outputs with timeSteps in graph are an artefact of model iterations
Parasitic losses, dominated by artificial liftDiscuss artificial lift parasitic lossesP Lift is the pressure against which an artificial lift system will work
Net electrical power after subtraction of parasitic lossesMedian power output around 3.5 MWCombinations of higher flow rates/and or unfavourable aquifer parameters show declining power outputs with timeSteps in graph are an artefact of model iterations
Impact of higher mass flow rates, in finite aquifers, on thermal and hence electrical power outputs300 days, no temperature drawdown, instantaneous power increases almost linearly with mass flow rate3,000 days temperature drawdown occurs at higher flow rates, therefore instantaneous power after 3,000 days declines at higher flow rates
An example of how long term simulations can be used to optimise a system is to estimate cumulative electricity generated after say 10 years.In this case, for mass flow rates above 30 kg/s, temperature drawdown will occur, so slope of graph slackens. Increasing flow rate above 30 kg/s gives diminishing increase in cumulative power
Optimisation is focused on optimising the power generated, typically number of kWh over a specified period.Controlled by uncertainty in external parametersNeed to select controllable parameters accordinglyIn simpler cases this can be done by spreadsheet, but dynamic system modelling tools such as GoldSim offer a more flexible and intuitive means of running scenariosCan be used at a variety of project stages