ELEC9714
Electricity Industry Planning + Economics
Assignment 2
Easier than the first assignment I think, certainly more constrained. Again, however, these are long and challenging assignments so marking is done accordingly. The key thing is that you understand the approach for each of the questions. As discussed previously, these assignments were excellent exam preparation.
Aim to make your assignment look like a professional consultancy report and paste in Excel plots and calculation tables.
I hope you tried to do this – good practice if you hope to work in the energy modelling and analysis space.
Note that late submissions can’t be accepted as the solutions guide needs to be released in time for the final exam.
I hope you got this in – one assignment where I can’t provide extensions due to the tight timing for the exam.
(2 marks per part for 20 marks total)
You are a market analyst working for a large international generation company considering investing into a wholesale spot market that has a mix of old brown coal plant, more recent OCGT, old but still highly reliable hydro, considerable wind generation and PV generation. This generation is mostly owned by four large generation participants as outlined in Table 1, which also provides annual capital and incremental variable costs for each plant. You may note a few similarities to the Victorian region of the Australian NEM.
Tried to capture some of the key aspects of Victoria which has ambitious renewables targets and aging brown coal plant. I didn’t add the interconnections as they would make it far more complex (would need to model NSW, Tas and SA markets).
The plants are of different ages and some have been refinanced. The fixed O&M costs of some of the old brown coal plants are also particularly high. Hence the different annual capital costs that don’t necessarily match the ‘new build’ capital costs of the different generation technologies. Each technology other than the wind and PV generation can be assumed to be entirely flexible (ie. it can be instantaneously started and shut down with no cost, and run at any operating level between 0MW and its rated output. It is also assumed to have constant incremental variable cost (ie. short run marginal cost - SRMC) over its entire operating range.
Key point is that even an old coal and gas plant still has annual capital costs unless it has been completely depreciated or written off. Even then, will still have fixed O&M costs. Keep in mind that these plants are still being bought and sold, and the banks who lend the money do expect to get their loans paid back. Sunk costs is still a relevant concept, however; regardless of what you paid for it a little operating profit is better than none at all so there are incentives to offer your generation into the market at variable costs. Note that the operating assumptions are pretty significant given that the coal plants have minimum operating levels, start up and shut down time requirements and costs. In practice, these plants would offer into the market with unit commitment / decommitment in mind.
Unfortunately the hydro plant is very energy constrained (ie. only able to operate at very low capacity factors of around 10% - hence it is offered into the market at $150/MWh. It therefore can be expected to run only occasionally but be highly profitable when it does so given that it has zero operating costs. This is a problematic approach but properly accounting for energy constrained hydro will make the assignment far too complex, and we do want our gas generators to make some money occasionally. Note that when the hydro is dispatched it is shared by Snowy and AGL according to the proportion of hydro they own – ie. 200MW of hydro dispatch goes 150MW Snowy and 50MW AGL; seems to be fairest.
In theory, for a simple market with only coal and gas plants, the gas peakers only make money when there is scarcity pricing – ie. demand isn’t being met and the price is set at the value of lost load (VOLL) or the Market Ceiling Price. This is of course a very low probability occurrence (perhaps only 10 hours or so a year) so I didn’t want to do that.
Instead, I use the energy constrained hydro to create some profit for the gas plants. Victorian hydro is certainly energy constrained so this seems reasonable. And remember how the gas peakers sell call options to underwrite their investment costs – well, energy constrained hydro sell calls as well. So it’sa reasonable fit.
Load is simply modelled as 9000MW during the day (7am-7pm) and 5000MW overnight (7pm-7am) – yes, a major assumption. There are a number of retailers but for simplicity you don’t need to consider them. However, there is one major industrial load, an aluminium smelter, which has fixed costs associated with debt servicing as well as fixed operating costs of $175,200/MW/year, and a short run marginal benefit (operating benefit) of $110/MWh. Note that its demand of 600MW is included in the total 5000MW and 9000MW during the day and night, and the plant operates at 100%CF over the year.
Thought about making this industrial load flexible and hence only operating when electricity costs mean it is making money. It does make the problem rather more complex and the smelters run 24/7 right now except under very unusual circumstances. They would rather get government subsidized electricity than work on the flexibility of their operation to turn production down/off when prices are too high.
Also thought about you modelling profits of the retailers but it involves more assumptions than for generation.
The total 3000MW of wind generation can be approximately categorized as operating at 80% capacity factor (CF) (300MW) for 20% of the time, and 30% CF for the rest of the time – its overall CF, therefore, is 40%. You can assume that all wind farms have completely correlated outputs – they all generate at either 80% or 30% CF at exactly the same time. Also, wind generation is entirely uncorrelated with day or night time. The 1000MW of utility PV generation can be modelled as a flat 600MW output during daylight hours – yes, another big assumption. All of the utility PV and 2000MW of the total 3000MW of wind are owned by a variety of smaller market participants.
In practice, far more complex probability distributions for wind and utility PV generation. Also, it won’t also all be correlated – that is, there are some wind farms in VIC that have rather different patterns of average wind generation to others which smooths out total generation (western VIC wind farms have rather different generation profiles from coastal VIC wind farms). Pretty much all of the VIC utility PV is going into the same region of the State and probably getting around 250-300 mostly sunny days / year. Wind and PV does make very complex to incorporate such effects – many market participants undertake these types of studies using Plexos, a commercial market dispatch and generation investment tool that costs around $40k/year.
Table 1: Generating Unit and Company Data for Question 1
Assumptions, assumptions assumptions – necessary to make the problem tractable without a proper market dispatch tool. For simplicity I gave all the coal and gas plants the same operating costs - in practice the differences between the coal plants and between the gas plants give some market participants a competitive advantage.
(a) For each of the two possible wind generation CF and two day or night scenarios (four possible scenarios in total), draw the generation offer curves on the same plot, and solve the spot market price. Assume that all generation participants are preference revealing (offer into the market at their operating costs) and note that the industrial load doesn’t bid to buy in the market but just operates at 100% CF while paying the spot price. The spot market price is set by the marginally dispatched generator. If demand coincides with a step in the supply curve then the price is set as the average of the two generator operating costs. Given the probability distribution of wind generation CFs and day/night time demand (and PV generation), calculate the time-weighted average market price. Put all answers in tables and discuss your findings.
A number of ways that you might bring these wind and day/night probability distributions together but the simplest is to just consider 4 market scenarios - high wind night, high wind day, low wind night and low wind day with probabilities over the year of 10%, 10%, 40% and 40% respectively.
Plot the generation offer curves and add the day and night demands (9GW and 5GW) to get market dispatch prices for each of the above scenarios ($20, $120, $20, and $150/MWh respectively). Average price is the weighted average of these - that is 0.1*20+0.1*120+0.4*20+0.4*150 = $82/MWh.
Interesting market - daytime prices depend on wind but not nighttime prices. Very large difference - much greater than actually seen in VIC… so far anyway. There is a lot of low operating cost coal in this mix and Victorian industrial demand has actually been falling. However, I ’ve used a daytime demand that is well above the typical VIC demand to drive high price outcomes. Actual daytime demand is getting reduced by rooftop PV, which we consider later….
(b) For each of the two possible wind generation CF and day/night scenarios, what is the total generation output (MW) of each generation unit for each generation company, the operating profit ($/hr) of each plant and hence company ($k/hour), as well as the operating profit of the Aluminum smelter … Put all answers in tables and discuss your findings.
Calculate the annual operating profit ($m/year) of each of the generation companies and the industrial customer given the wind CF and day/night probability distributions. Put all answers in tables and discuss your findings.
Solve market dispatch for each generating unit and the smelter for each of the four wind/day-night scenarios given the market prices. As discussed in the assignment consultations, it isn’t clear how to split dispatch of coal or gas when it’s setting the marginal price - it doesn’t matter for profits since if your plant is on the margin then the price you receiveis your operating cost. For the hydro, however, the split of dispatch is critical because even on the margin they are actually making excellent money. Hence, as I noted, you should split hydro dispatch according to ownership - 75% of any dispatch to Snowy, and 25% to AGL reflecting their 1500MW and 500MW of capacity respectively.
All generation technologies make operating profits in this markets. That coal plant is pretty profitable isn ’t it given its low operating costs - the gas less so. Wind and PV are doing pretty well - particularly PV. Interesting to note that wind makes money both at times of low and high wind - low wind has higher prices while high wind times have lots of MWh. Overall, the generators seem to be doing pretty well - operating revenues (ie. income) 3 X greater than their operating costs. The smelter also is doing pretty nicely as well despite the fact that doesn’t actually make an operating profit for 50% of the time (ie. when price is greater than $110/MWh). It really would do better if it was able to operate more flexibly.
Note that operating profit is a key issue in electricity market design; if a participant doesn’t make an operating profit then it is trouble, and there is a chance that the market will lose that generation (and perhaps even load). If a participant is making an operating profit but not an overall profit (see below) then that is more a problem for the bank/equity that provided the money for building/buying the portfolios. Participant still likely to remain in the market because some profit is better than none.
(c) Now calculate the total profit ($m/year) of each of the generation companies and the smelter given the wind CF probability distribution ($m/year) after covering the fixed (capital and fixed O&M) cost repayments that they are required to make. Also calculate these annual profits as a % of total fixed costs for each market participant. Please put your answers in a table. Discuss your findings.
Ok - once you account for fixed costs (capital and fixed o&m) then overall (net) profits for the generators and the smelter fall significantly. Still, all the generators are actually profitable even after covering these costs. Will often see profits reported as a % of fixed costs (sometimes also total costs). Those brown coal plants are cash cows with very high profits as % of fixed costs. PV is very profitable. Wind and gas are both providing reasonable asset returns. In this regard, AGL is doing very nicely from Loy Yang. Overall, the generators are doing pretty well with the exception of Snowy which doesn’t have that brown coal plant. Profits of 41% of fixed costs is pretty attractive - surely some attractive investment options.
(d) Your international generation company is considering investing in a wind farm in this market. You are aware that there are other international companies that are also contemplating a similar investment and it seems entirely possible that 2000MW of new wind will enter the market over the next couple of years. This wind generation will be almost entirely correlated with the existing wind, however, improvements in wind turbine technology will deliver a higher capacity factor of 50% through better low wind performance
- 70% CF for 20% of the time but now 45% CF for the remaining 80% of the time. Given capital costs of $1836/kW for these wind projects, with financing available at 6% over 20 years, and fixed O&M costs of $20/kW/year, the same operating costs as other wind of $5/MWh, estimate what price ($/MWh) you could sell a 20 year PPA (a variable volume CFD around the future wholesale market price) for. Do you consider it likely a retailer or the industrial customer might be interested to purchase this PPA given current market prices?
Common situation – ifyou ’re interested in making an investment there’sa reasonable chance there are other players considering the same. Means there’s a good chance that there will be more new capacity entering the market than what you propose to build yourself. And as we ’ll see, this impacts on future prices in potentially problematic ways for investors. Newer turbines so better performance than the existing fleet.
Use the Capital Recovery Factor equation you first encountered in Assignment 1 to calculate a yearly capital cost ($160k/MW/year). Add the fixed O&M and then divide by the expected annual generation (50%CF X 8760 hours) to get $41.1/MWh to cover fixed costs. $5/MWh variable O&M costs gives a levelized cost of around $46/MWh. Could certainly see if can get a better price but you will cover costs at a PPA of $46/MWh. Certainly a starting point for the discussion with potential buyers, although you wouldn’t likely let them know your minimum acceptable price.
Given an average market price of $82/MWh at present you ’d reckon that retailers and industrial loads like the smelter might be pretty interested to buy a PPA for less than $50/MWh. But remember that this new wind won’t earn the average spot price given it’s output distribution - it generates more of its output at times of lower prices during the day. Still, if you assume that market prices don ’t change then buying a 100MW PPA at $46.1/MWh would cost you 100*8760*0.5CF*46.1 = around $22m/year but the electricity you get from the wind farm would be worth $35.9m/year so you’d do very nicely. Effectively the wind farm earns $80.80/MWh in the market, not much less than the average price. However, can you assume that market prices don’tchange?
(e) There is of course the question of all the other wind generation projects, and whether they will proceed. For the case where 2000MW of new wind generation does get built by you and other developers (not any of the existing companies), without any PPAs (ie. all the developers take full exposure to the market price), estimate the total profit or loss made by the 2000MW of new wind ($m/year and % profit/fixed costs). You will of course need to recalculate spot market prices and generator dispatch for each of the four combined wind CF and day/night scenarios (an additional 1400MW of wind for that 20% of the time and 900MW of wind for the other 80% of the time. Is there a commercial case for building this wind? Also estimate the impact on overall profitability of all the other generator companies and the industrial customer. Be sure to put your results in tables and discuss your findings. In particular, what does it say about the value of the PPA to whoever bought it.
That much wind does change pricing - a lot. We now have 1400MW of additional wind generation in the high wind scenarios, 900MW in the low wind scenarios. New dispatch curves show price is $20/MWh for all times except when there is lower wind during the day when it is $120/MWh. Profitability of all generation participants takes a hit. Average price is now $60/MWh. Interestingly, the new wind actually makes a reasonable profit, earns an average $56/MWh, well above its levelized cost of supplying. However, hydro and gas no longer earn any operating profit and the coal plant can’t make up for this. All generation technologies other than PV and the new wind now making a loss after fixed costs. The future for the gas plant now looks potentially rather bleak given they don’t even make operating profit, while the brown coal plant is now worth a lot less - the owners would have to write down the value of their assets, any future sales prices would be adversely impacted. The hydro will just wait it out -fixed costs aren’t too high and they will eventually get back into the money once the gas and/or coal leaves. The utility PV is still highly profitable, but the existing wind farms can no longer cover their fixed costs (although still providing a pretty good operating profit). Keep in mind that the worst thing that can happen is that you can’t even make an operating profit - that’s when you start contemplating closing the plant. Strictly you should distinguish between fixed o&m versus fixed capital costs in questions of profitability - if you can ’t even cover your fixed o&m then you really are in trouble. Not being able to cover your capital cost repayments is the banks problem (or the owners if they funded the project themselves off their balance sheet). In summary, a reasonable investment case for new wind despite it’s impacts on all the other participants. Might expect that the incumbents won’t necessarily be so keen to see more wind come in. The smelter is in a wonderful position how with significantly improved profitability from the lower prices. We have actually seen this in Australia as renewables investment has contributed to pulling down wholesale market prices. Some of the major industrial energy consumers have gone from opposing renewables because they ‘increase’ prices to welcoming them for ‘reducing’ prices. Could certainly envisage retailers and industrial loads looking to sign PPAs secure in the knowledge that if all that wind doesn ’t come (they won ’t contract the full 2000MW themselves in all likelihood) then they’ll do really well out of the PPA, and if all that 2000MW of wind does come they ’ll still do well from both the PPA but also more generally from the lower prices.