HOME

OZ-ENERGY-ANALYSIS.ORG   -   open science for the new millennium

THE STORIES   |   DATA   |   ANALYSIS   |   MODELS   |   LITERATURE   |   DISCUSSIONS

status: [16th March 2011] round one

Fifty percent Renewable Scenarios


ABSTRACT

OzEA now starts considering explicit scenarios for 50% renewable supply in the electricity sector. Early scenarios are rudimentary, and will be refined in an iterative way. For now the focus is on 2003, as dictated by the current solar data. Of particular interest are the demand remainder characteristics.

The task is to site, and assign capacities to, both Concentrating Solar Thermal (CST) plant, and wind farms. Siting is currently limited to those places where we have established wind speed data or solar radiation data respectively.

With the overall NEM demand at ~24 GW average, we seek ~12 GW from solar and wind to obtain around one half of the national electricity requirement from renewables.


INTRODUCTION

By establishing a starting-point 50% renewable scenario we can proceed to examine the consequences and refine the scenario. This in itself is expected to be an ongoing process into the medium term. The reason we need some strong 50% scenarios is so that 30% and 40% scenarios can be built, thus establishing an evolutionary pathway to simulate debate and discussion, and as a basis for further work.

This page is limited to establishing the scenarios (plant and transmission lines), with some associated analysis and collation of results about these scenarios. More detailed work such as establishing flows in transmission lines, and the necessary output from fuelled generators, will occur on other dedicated pages. In the initial stages here the transmission aspects are to be ignored.

The solar data in particular dictates focusing on the years 2003-4-5, and first we work on 2003. Thus there is no demand data for Tasmania (in addition to WA and the NT). The overall demand for the NEM states in 2003 (QLD, NSW, VIC, SA) averages to around 20.6 GW, and while we work below (as per abstract) to set out 12 GW average of renewable supply, this is scaled back to 10.3 when constructing demand remainders.

map showing first 50% renewables scenario
map (being developed) showing the first 50% renewables scenario

The map figure above shows the renewable supply configuration established for the first 50% renewable scenario, and is followed below by some comments and analysis leading to specifying capacities. Click on the map to get the PDF, and then zoom in and pan around. This will become more important as more detail is included (e.g. transmission lines). Further down again is the demand remainder analysis, including a first gentle prod at the question of storage requirements. I then make some comments to round out the head post, including on the east-west-transmission-link question, before it all turns over to the discussion comments, where developments and evolution mostly happen.


METHODS and DATA

Hypothetical plant is constructed by the application of a wind power curve / CST power curve model to the wind speed data / solar radiation data to obtain output traces for simulated wind farms and simulated CST plants (hourly / half-hourly time series respectively). While these underlying aspects can be developed and refined in various ways, we proceed with what we have for now.

Using the available simulated renewable plant (above), we seek to choose capacity values (sizes) for the individual sites so that the overall combined renewable supply behaves in desired ways. That is we seek to shape the demand remainder profile, and it is critical the reader understand both the 'demand remainder' (being full demand minus renewable supply), and the demand remainder profile (see link in abstract).

In the first instance we simply seek to reduce the maximum value of the demand remainder, and use a simulated annealing approach to do this (as detailed in the comments).


RESULTS

[25th March 2011: The reader can safely skim the next two sections (The CST solar farms and The wind farms), while taking care to follow The Demand Remainder section.]

The CST solar farms

Using the simulated CST farms (in particular this data for 2003) we examine how the available sites might contribute to a national electricity system. This is necessarily crude, and the end point for here and now is simply to select sites and assign capacities.

To start, the simulated CST farms are set at 100 MW capacity by observing that 250 W/m2 intercepted is the usual peak level obtained under the current CST power curve model (with occasional incursions above this when the direct irradiance holds at > ~1100 W/m2). Thus, a collecting area of 400,000 m2 would be required to achieve 100 MW in strong direct sun, and we use this as a starting point for considering all 15 CST data sites. This 100 MW is akin to nameplate capacity for a wind farm - the average output is only a fraction of the peak (i.e. dark half the time, not always strong sun other times).

In order to get a foothold into future analysis, have used a monetary approach to examining the simulated CST farm output. That is, have taken the price data for the NEM states for 2003 (SA, VIC, NSW & QLD), and have used this to calculate the monetary return the CST plants would have made if operating at this time and receiving those prices. This is shown in the table below (with WA receiving SA prices, TAS receiving VIC prices, and the NT receiving QLD prices). Also calculated is the return based on a flat rate of $27 per MWh (the market average for that year, and a low value compared to other years), and using an overall - but time varying - NEM price.

The 'NaN days' column gives the number of days that contained NaN (missing data) runs, these being substantial in some cases, and with missing data requring careful handling. For the Regional and NEM price calculations, have also given the percentage of the overall return that occurs in the most lucrative one percent of the time. 

   site:    NaN   CST_avg   flat_rate    Regional   top 1%   NEM_price  top 1%
            days   (MW)    (mil OzEA$)  (mil OzEA$)         (mil OzEA$)
  ---------------------------------------------------------------------------
   ADEL      76     17         4.0            4.8     19          3.3     9
   MTGAM     54     15         3.4            4.4     19          3.4    20
   CRNS      14     18         4.1            3.3     13          3.9    18
   ROCKH     25     21         4.8            3.6     10          4.5    17
   WAGGA     3      21         4.9            4.6     17          4.5    12
   MILD      30     23         5.3            4.8     10          4.8    13
   MELB      4      15         3.4            3.2     12          3.2    15
   CGRIM     13     13         3.0            2.7      9          2.9    16
   BRME      4      28         6.5            8.4     15          8.4    33
   LRMTH     59     29         6.7            8.1     14          8.3    29
   GRLDN     18     24         5.6            7.8     18          6.3    25
   KALG      36     24         5.6            7.4     19          5.9    26
   DRWN      32     19         4.3            4.6     33          5.2    32
   TENNT     27     29         6.6            5.5     20          6.5    21
   ALICE     14     29         6.8            5.4     12          6.5    17

Note the use of 'OzEA$' rather than '2003AU$' - will do this often to signify that the monetary values need to be carefully understood before being lifted into other contexts.

We see that the better sites give up to 29% nameplate as the average, and the poor sites (e.g. Cape Grim in Tasmania, Melbourne) are below 20%. What is also interesting is that there is no clear increase in return at the variable market rate/s compared to the flat rate, excepting the Broome and Learmonth sites, suggesting that perhaps some modest thermal storage as part of the CST might allow better exploitation of the early evening market.

What remains is to select from this limited menu around 6 GW average of Concentrating Solar Thermal electricity generation:

               ADEL  MTGAM  CRNS  ROCKH  WAGGA   MILD  MELB CGRIM  BRME  LRMTH  GRLDN  KALG  DRWN TENNT  ALICE
  supply (avg)   -      17    -    1050   1050   1150    -     13    28     29     24  1200    -     29   1450
  nameplate      -     100    -    5000   5000   5000    -    100   100    100    100  5000    -    100   5000

Here I have taken five good sites to have large (very large) 5 GW nameplate CST solar farms, and have a further six sites kept in the analysis with 'token' 100 MW plants. In terms of average supply, this gives: 5900 + 145 = 6045 MW, as required. With a further ~6GW from wind, have around one half of the national electricity requirement from renewables.


The wind farms

As detailed here we have 30 simulated wind farms established for 2003. Pricing the electricity from the individual simulated wind farms turns out not to be especially interesting or helpful. As a blunt average ($27 a MWh), a 100 MW nameplate wind farm with 35% capacity factor will earn 8.2 mil OzEA$ for the 2003 year, while pricing the 30 simulated wind farms by the NEM price (weighted average of regional prices), gives returns mostly in the 7-8 million range, with Goulburn Airport (BoM_070330) being the only one to beat the flat rate, coming in at 8.9 million -- and with two SA sites taking up the rear with 6.3 and 6.5 million (BoM_023373 and BoM_023875).

We wish to have a total average of 6 GW wind power, and this can be achieved by simply having each of the 30 wind farms at 570 MW nameplate capacity (as the wind farms have an imposed 35% capacity factor: 30 * 0.35 * 570 = 5985 ~ 6 GW, as reqd).


The Demand Remainder

For the above described 50% renewable scenario, for analysis of the year 2003, with the renewable electricity from wind and CST in equal parts, the following demand remainder profile is calculated:

demand remainder profile

erratum: the plot should say 25% CST and 25% Wind
(adding to 50% renewable electricity)


Of immediate interest here is where the tails end. The raw demand tops out at 28.5 and the demand remainder for our scenario has toped at 24.1 (a fifteen percent reduction; i.e. a 15% capacity credit for 50% renewables). Ideally this wants to be somewhere ~50%, although what really matters is a more complex calculation of costs for peaking infrastructure. If this can be bought to ~50% without needing to do cartwheels, that will be very encouraging and constitute a significant win in addressing the problems, real and perceived, with high penetration renewables. If, conversely, this empirically calculated capacity credit remains stubbornly low, then it's not the end of the world, but nor is it encouraging.

[Fri 25th March 2011]
There has been some confusion here about the use of "capacity credit". The antidote is to properly understand what the demand remainder profile represents (see link in Abstract). Note also that we do not use, nor rely upon, the "capacity credit" for any calculation or conclusion - its use is an artifice with two specific purposes: (i) to get started on shaping demand remainder profiles via the parameterisation of the renewable supply, and (ii) to (hopefully) score a rhetorical point in dismantling the "renewables can't do baseload" [sic] meme.

Storage will likely be the key here, both centralised (as Pumped Storage Hydro, also maybe some boutique hydrogen), and internal to the CST plant. To start we use a centralised view of storage, eventually moving onto the economics of CST-site-storage vs. transmission and networked storage.


Transmission

The transmission lines are not so interesting yet. Can (and will) draw lines on the map connecting the generators to loads, but before much can be said about these it seems necessary to bring in the machinery for (a) defining the flows, and (b) pricing the electricity and thus moving into the economics. These aspects are being worked on, but it all takes time. Sun goes up, sun goes down, arms on the clock go round and around.



(Opening) DISCUSSION

A first concrete scenario, however rudimentary, provides a national renewable supply trace that can be examined and refined. Even before needing to address the transmission requirements, the gross demand remainder profile gives insight and a specific problem to work on. That is, how can the scenario be adjusted in order to shape the demand remainder.

For now we include transmission capability as needed to connect the renewable supply to loads, including a large east-west transmission capability. In time will need to explicitly address the pros and cons (and likelihood, timing, sizing) of any major new transmission infrastructure required by the scenarios.

OzEA now starts including costs into the analysis and modelling work, albeit just around the edges for the moment. One very particular reason is that in optimising system configurations it is ultimately overall cost we want to optimise against. However, please be very careful not to take cost values from these pages to other contexts without a full understanding of what they embody.

Expect the latter part of 2011 will involve detailed modelling of the electricity flows and the temporal requirements from the intermediate and peaking supply infrastructure. This will also involve modelling the market (i.e. the electricity price). This work will of course be developed openly and incrimentally.




DISCUSSION and DEVELOPMENT:

1

OzEA_MFPCRS0001

Neil Howes
Subject: 50% renewablesl demand remainder
Date: 2011-03-20 (at 19:36:41)

50% renewable scenario would be replacing 12GW of coal fired power, or about 16GW capacity(assuming a 75% capacity factor). Existing 6GW of hydro and 7GW of OCGT capacity( the most flexible and lowest CO2 emitting sources) would still be available, along with a residual of 12GW coal-fired capacity, giving 25GW available to cover the 24.1GW demand remainder.
If a small amount of CSP thermal storage can reduce the demand remainder as seems likely that would give a comfortable reserve. CSP storage has the additional benefit of making better use of CSP turbine, collector and transmission capacity. Another solution would be to increase the number of turbines at TAS hydro sites ( along with beefed up BASS link) and use TAS hydro at a lower capacity factor.
If the high demand remainder of CSP is due to prolonged periods of wide-spread cloud the solution would be to retain some of the 16GW coal fired capacity on standby during high cloud cover or high demand periods(summer heatwaves) but use at a very low annual capacity factor.

2

OzEA_MFPCRS0002

francis
Subject: wheel kicking the 50% renewables demand remainder
Date: 2011-03-22 (at 17:12:29)

Thanks Neil, that seems a solid take of how a 50% scenario could work, and using the coal stations on a lesser schedule and as backup can be part of the transition to 30 and 40%, but is not really something OzEA wants to have too much of in the 50% scenarios.

While the storage aspects are to be developed, some comment can be made now. There is around 20 GWh of Pumped Storage Hydro (PSH) to include, and usually a much larger amount of run-of-river hydro available. Because the run-of-river hydro can become limited on occasion, it can not be relied upon (but will in practice be used). There will be some trade-offs between the PSH and CST storage, although I suspect that CST storage will be a way of introducing more storage capacity into the system, at a cost, but including the benefits you mention.

Shortly will get onto some 'wheel kicking' the demand remainder.

While seeking a 50% contribution to the capacity-credit is somewhat artificial, I also reckon it an important exercise. The first approach will be to shake up the weightings of the renewable generators to see what changes can be induced in the demand remainder profile. In addition to looking at the capacity-credit question (i.e. the x-axis end point), there is also the 'y-axis' intercept and the 'tipping point' as seen in the raw and 50% CST curves on the demand remainder curves in the head post. Understanding how flexible these profiles can be with judicious selection of sites and inclusion of storage is my focus here for the next little while.

3

OzEA_MFPCRS0003

francis
Subject: shaping the demand remainder using a simulated annealing approach to set site capacities
Date: 2011-03-23 (at 19:19:49)

Have been developing a simulated annealing implementation that shakes down the site capacity values in order to push the shape of the demand remainder around in a desired way. For now have been simply using the peak point of the demand remainder as the Objective. Two minor results to report:

1. constraining the problem to require 25% from Wind and 25% from the CST sites, the peak can be reduced to 22.2 GW (down from the initial 24.1, and compared to 28.5 for the raw demand). At this point I do not report on the capacities of the particular sites other than to note that the CST goes west (as expected). The 'improved' demand remainder is:




2. Without constraining the division between the CST and the wind (simply requiring 50% overall from both), the peak of the demand remainder can be reduced further to 20.2 GW, and with the wind dominating at 44% (CST 6%), to give:




These results are of themselves not particularly important. What is important is the development of machinery for pushing and pulling a demand remainder in desired directions, and my next post will detail this machinery (this may take another couple of days...) After that we examine some basic storage dynamics.

Please note that I do not claim any particular optimality for this procedure, but simply that movement in a desired direction has patently occurred.

4

OzEA_MFPCRS0004

Ben McMillan
Subject: The overall picture
Date: 2011-03-24 (at 09:15:03)

This is a very interesting bit of work.. in the end choosing levels of storage and wind vs CST or PV or even geothermal is going to have to look at tradeoffs of the shape of the demand remainder and price.

Playing with the demand remainder by itself is an informative exercise, but it is worth keeping in the back of the mind the overall power system.

At 50% renewables, one is considering a system where renewables and fossil plants work together. It seems to me that you need to know what the aims of the system design are (reduced capital and carbon costs?) and how the non-renewable generators will work if you want to optimise the overall outcome (storage versus dispatchable generation?).

The approach here, which is to work towards 50% capacity credit as well as 50% of generation, is almost as stringent as requiring the renewables to generate 50% of the power at each time point (which would, in a sense, be a half-sized 100% penetration system). Sort of artificially putting a firewall between the renewable and nonrenewable system.

5

OzEA_MFPCRS0005

francis
Subject: capacity credits
Date: 2011-03-27 (at 02:31:24)

Ben, it almost seems like a pulpit, for a lesson on capacity credits.

The point of bashing at them here, in addition to being trainer wheels for demand-remainder-shaping, is to burry them as a way of judging the prospects of high penetration renewables.

In a real sense the empirical approach here actually calculates the "capacity credit" (after considering many years of data), rather than being a wooly statistic. The idea that a 50% renewable system can be doubled into a 100% system seems similarly wooly. But none of it really doesn't matter; the job is to keep at building the 50% scenario, first this bit, then with transmission and a go at the market. With the spatial model established (all 2011?), can pin in 30% and 40% scenarios with more discussion on timeframes and dollars. In all of this we also need models for the fuelled supply, so that for everything we can define it's behaviour in the market and get some useful dollar values. I continue to suspect that individual/household/node participation in demand management, and hydrogen, have some interesting timeline issues, and might hold some surprises (at least to me).

6

OzEA_MFPCRS0006

francis
Subject: machinery worksheet
Date: 2011-03-29 (at 00:18:37)

The 'v0' machinery for annealing site capacity values so as to reduce the peak value of the demand remainder is now detailed in this worksheet [next version of worksheet]. Running the machinery and obtaining capacity values for sites continues to be exploratory. Here is some example output, which may be taken as underpinning the figures in #3 above.

As previously, two analyses are performed: one where the CST and Wind are constrained to each supply 25% of the overall electricity demand, and a second where no division is imposed and we simply require that the combined CST and Wind supply 50% of the overall electricity demand.

In the constrained case, the CST sites are dominated by Learmonth in particular, in an expected flight to the west caused (presumably) by the peak demand often occurring in the early evening. Storage and other considerations can be expected to alter this dynamic, and so for now the CST breakdown is considered uninteresting.

For the unconstrained case, there is a move towards wind (~44%) over CST (~6%), again reflecting the consistent mismatch between when the CST is producing power, and when power is most needed.

For both cases, the wind sites, in combination with the CST, arrange themselves into interrelated, but not always the same, patterns. There are however some consistent 'good' and 'poor' sites (see file). Be aware, however, that chance relationships between peak times and the output from these sites may be driving the observations. It will be interesting to see if these same sites come up again with analysis of the 2004 and 2005 data.

Finally, a broader caveat. This work is exploratory, with two purposes: first, we are seeking to reduce the peak of the demand remainder for both practical and rhetorical purposes, and will next proceed directly with this aspect. Also, the machinery being developed for shaping demand-remainder-profiles may be conceptually straightforward, however, the practical issues with coding and understanding just what is happening are more complex -- much of the effort being expended is aimed at developing understanding of what this machinery can do, and how best to configure it to achieve smooth use.

7

OzEA_MFPCRS0007

francis
Subject: the DRS curve + 'Capacity Credit' sign off
Date: 2011-04-04 (at 16:20:10)

The DRS curve is the Demand Remainder after Storage (green curve in plots following).

Based roughly on a doubling of the existing Pumped Storage Hydro (see PSH link in #2), have constructed a simple storage model with 40 GWh capacity that can provide up to 5 GW of supply, and can 'charge' at 2 GW (and only 2GW, and for a minimum of 3 hr). To start, I schedule this storage in a simple way, seeking only to shave back the demand / remainder above some threshold. With the current Demand Remainder (see #3 above) topping out at 20.2 GW (the 'TOP'), we can potentially come back as far as 15.2 GW, although the total capacity and charging constraints also apply.

After a little experimentation with parameters I have found it straightforward to have the DRS peak at ~16 GW, as shown:

DRS with blunt nose

Here the storage algorithm (applied within the objective function of the annealing) has sought to supply any demand in excess of 78% of the TOP; whenever the demand is less that 68% of TOP for at least three hours the storage goes into 'charge mode'. Note that the power for charging the PSH will come from both the renewable and non-renewable generators (as it does now).

In practice, the storage would be used more intelligently to flatten out the daily peak (so long as the storage is able to recharge sufficently for any larger peaks in the days ahead), and such scheduling gives a DRS curve with a more gradual approach to the peak. To include this in a simple way I use the same approach as above, but with TOP continuously reevaluated as a local parameter within a window of some number of days ahead. Looking ahead one week, the following DRS is achieved:

DRS with tail

In the above, the following generators dominate (with given nameplate capacities):

CST:
5.6 GW Broome
3.8 GW Tennant Creek

Wind:
3.4 GW at BoM_033295 - QLD, ALVA BEACH
2.7 GW at BoM_039322 - QLD, RUNDLE ISLAND
2.5 GW at BoM_088051 - VIC, REDESDALE
2.4 GW at BoM_009542 - WA, ESPERANCE
1.8 GW at BoM_008051 - WA, GERALDTON
1.6 GW at BoM_041359 - QLD, OAKEY
1.4 GW at BoM_069137 - NSW, GREEN CAPE
1.2 GW at BoM_018195 - SA, MINNIPA
and:
500-1000 MW from: BoM_069138, BoM_026095, BoM_026021, BoM_009215 and BoM_097080
100-500 MW from: BoM_056238, BoM_090186, BoM_022046 and BoM_040764

At this stage these generators are simply noted for future reference. There are (at least) three aspects that prevent these sites being interpreted as desirable for development: first, with only the 2003 data so far in play it might be pure chance that a renewable resource is running strong at the same time as demand is high; second, the annealing algorithm used does not identify some unique optimum, but rather one or another interrelated configurations of generators that together act to reduce the peak; and third, it is to be expected that spatial treatment of the problem (i.e. with transmission and localised loads) will alter the dynamics.

Note also that the DRS curve here, peaking at ~16 compared to 28.5, represents a nominal capacity credit of 44% for 50% renewables. As in the head post, I expected it would be easy enough to achieve a result of this sort, and stuck my neck out in anticipation. It can reasonably be argued that this 44% figure is not, or has not been shown to be, robust - and so no strong claims are made. However, those who wish to argue against high penetration renewables on the basis of capacity credit considerations can now consider the onus on them to demonstrate their case. Here we will continue, incrementally, to develop understanding and intelligent configurations of 50% renewable electricity in Australia.

8

OzEA_MFPCRS0008

neil howes
Subject: 50% renewable role of hydro
Date: 2011-04-05 (at 06:47:30)

Francis,
This is great work you are doing. If you are projecting replacing 50% of present power generation(ie 65% Coal fired) you should also include the existing reliable hydro( in addition to that used for pumped storage). This would be about4GW.

9

OzEA_MFPCRS0009

Ben McMillan
Subject: Robustness, corrections
Date: 2011-04-05 (at 17:31:21)

I agree that this is great work: I was pessimistic that you could get so close to 50% capacity factor.

You have already mentioned checking whether the same optimised system works well for different years (on data the optimisation algorithm hasn't seen).

Another methodological query is whether the algorithm has selected less realistic 'outlier' wind farms: for example, we know that the diurnal variation of the wind farms is difficult to predict from the station data, and this will be important in the optimisation weighting.

It would be interesting to see how well you can do with all the wind farm sizes the same. There are other constraints apart from demand matching on wind farm locations and sizes, so this might also be more 'realistic' or conservative.

10

OzEA_MFPCRS0010

francis
Subject: Re: [neil howes] 50% renewable role of hydro
Date: 2011-04-08 (at 12:09:49)

Of course we still need to get to the transmission issues, but for now inclusion of 4 GW of "reliable hydro" would allow further reduction in the peak of the demand remainder.

As the hydro resource is so large, I suppose that IF use were limited to the more extreme peaking events, then it would be much easier to guarantee a certain level of supply through low-rain periods. Is this what you mean by "reliable"? Would be great if we can flesh this out a little (ideally finishing with a summary on the generators page), and once this is clear I'll implement and we can see how much this contributes.

11

OzEA_MFPCRS0011

francis
Subject: Re: [Ben McMillan] Robustness, corrections
Date: 2011-04-08 (at 12:12:09)

Robustness is important, eventually critical, but not yet a priority. I agree there are issues with the diurnal variation (see also this comment), and also with solar capture geometry (and also with data, and especially the lack of solar data for central Queensland).

You say "there are other constraints apart from demand matching on wind farm locations and sizes" -- do you have in mind anything specific here?

Not clear to me what it would achieve to do a storage model run against all simulated wind farms at equal capacity (same for a selection of the CST sites), but nor is it an onerous task. I also have in mind to do some runs that exclude all WA sites; SO, perhaps there are some other 'reference' cases that it will be useful to consider? I'll get to processing these and posting before too long.

12

OzEA_MFPCRS0012

Ben McMillan
Subject: Constraints
Date: 2011-04-13 (at 08:59:15)

The main considerations for wind farm sites are usually wind resource quality and cost of transmission (and legislative approval!). These have a big influence on cost/MWh, so have to date been the main criteria for choosing where wind farms get built.

So I wouldn't be surprised if a real-world system of wind farms wasn't at all optimised for improving the capacity credit. It would be interesting to know how much difference that makes.

Solar is a slightly different case, because taking advantage of e.g. late afternoon sun in WA might change the economics substantially.

13

OzEA_MFPCRS0013

francis
Subject: Storage models and revised worksheet
Date: 2011-04-14 (at 12:56:32)

Have now implemented a model for thermal storage at the CST plants, and otherwise done a significant round of tidy-up on the codes [revised worksheet v1 here]. Note that this worksheet also includes the code for centralised storage, as described and used above. What follows is a description of how the thermal storage model works; results showing use of the model to follow in another post.

[while presenting the worksheets is time consuming, and unlikely to get much attention, the Open Science model requires that I keep these in order]

Consider first a very simple storage model, where the total input for the day is levelled and extended into the evening some number of hours; this would simply require that (a) the total power input for the day equals the total power output over the extended period, and that (b) the cumulative output (in the initial hours) does not exceed power collected to that point. This model ignores losses, as does the refined version of this basic model described in what follows.

The above approach is not responsive to the actual demand needs, whereas a CST plant operator would regulate output (to the extent possible) in order to achieve the best price, which for now we simply take as preferentially providing power at higher demand times. This is achieved by use of a simple heuristic, which may later require refinement both to improve performance and to better reflect engineering realities.

Heuristic is to first establish three time points (for each day), t1, t2, and t3, representing the time of first generation / solar input, the time of last solar input, and the time to which the plant can keep providing power from thermal storage, respectively. Explicitly, t3 = t2 + S_hrs. Note also that 'days' are here taken on a 4am to 4am cycle. Within the [t1,t3] time window, we seek to provide output in a way that is weighted to follow demand. This 'demand' is simply what is provided to the function, and it is anticipated to be the demand remainder generated by subtracting the wind supply from raw demand. It is the weighting process that is the core of the heuristic.

The given demand is transformed into 'demp' values by squaring the demand in order to give increased weighting to the higher values. Using these 'demp' values, and with the total power harvested for the day as CST_tot, the output at time t is taken as: Pout(t) = demp(t) / demp_tot * CST_tot, subject to two constraints. The first constraint, as above, is that power cannot be given out before it has been collected, and second, that the output does not exceed 110% of the nameplate capacity of the CST plant. The codes in the worksheet formalise this procedure.

14

OzEA_MFPCRS0014

francis
Subject: model visualisation
Date: 2011-04-15 (at 11:55:37)

For looking at these models, here is draft one of a temporal graphical view. The images below are just thumbnails -- you need to click through to the pdf, zoom in, and pan along.

Shown is the basic 2003 model given in the head post (no shaking down of site capacities), with the two storage models included.
The blue trace is the historical demand, with (modelled) charging of centralised (PSH) storage added in solid blue.
The light-blue fill is the power from the wind farms being directly subtracted from the demand.
Along the bottom an orange trace show the daily solar power collect, which is acted upon by the CST storage model (nominally 6 hrs storage) to give the yellow fill that is further subtracted from the demand to give the demand remainder (DR: solid back trace).
Release from the modelled PSH is shown in magenta (DRS: Demand Remainder after Storage), further reducing the demand that remains to be met by the fuelled 50% of overall supply.

model view part 1
model view part 2
model view part 3
model view part 4

I am pleased with these plots, and the issues they show up. They will in time guide us to (a) a good 50% scenario, and (b) a good understanding of the changes to be implemented through 30% and 40%.

While all the tedious coding machinery is needed for this work, using a visualisation approach of this sort will help both in driving the machinery and presenting in a more accessible way the path that we drive down.

note: fc back to holidays now.

15

OzEA_MFPCRS0015

francis
Subject: centralised storage (PSH) model codes
Date: 2011-05-17 (at 09:58:14)

The code for simulating centralised storage (considered as Pumped Storage Hydo) has developed from that in #7 above. The current improved version (v2) is here: OzEA_SASC_central_storage_model.m This is what generates the PHS used in the figures in The Third Story. The role of the PHS is not just to shave peaks - it is also to provide demand when the demand remainder is low (thus lifting the base), and this aspect requires further work (expect to do a further round of development on this heuristic before too long).

[Update 4pm] I forgot to say that from now until otherwise stated, OzEA is working with 30 GWh of PHS (an increase from the existing ~20 GWh). The model can provide supply between 0-4 GW (if not exhausted), and charging occurs at (and only at) 1.5 GW for a minimum of 3 hr. See code for more details.

16

OzEA_MFPCRS0016

francis
Subject: taking stock
Date: 2011-06-02 (at 15:09:43)

The focus here was on getting material for the third story. That's now done.

The tentative timeline is to establish for ongoing work a round one "50% scenario" by early September. A big sub-task will be an analysis of major transmission. Tasks for following up here (in July) include:

- some examination of how the storage parameterisation (CST storage capacity and PHS capacity) effects dynamics
- elimination of sites that do not make a particular contribution (i.e. simplification)
- consideration of the 2004 and 2005 data
- bringing in a cost based Objective for the annealing of site selection

Other aspects that will require consideration include:

- the role of PV (perhaps contributing as much as 10% toward the 50%)
- inclusion of pure hydro (there is some 7.8 GW of installed capacity)
- having coal-based power (in the future) vary with the low-frequency aspect of the demand remainder.


[Jan 2012 update] whole project fell over in July 2011. I'm hopeful of being able to bring significant focus back to this work (OzEA in general) from Feb / March.

[Show Full Lists]

Post Comment:

A name or alias, email and concise subject are required. Your email will not be abused.
Comments are required to be polite and on topic (commenting etiquette)

Name:*
Email:*
Website:
Subject:*
What is the longhand for 'Oz'? :*


HOME  

  TOP   |   REFRESH PAGE   |   MODELS

fc - 17th Mar 2011