Solar < (cheaper than) Coal Power in Emerging Markets like India

... and (much) faster than you think

Shivkumar Kalyanaraman

Twitter: @shivkuma_k, 

Email: shivkuma@gmail.com or shivkumar-k@in.ibm.com

{Views are Personal. ShoutOut to collaborators: Sukanya Randhawa, Vikas Chandan, Pratyush Kumar, Karthik Visweswaraiah @IBMResearch - India, Rajesh Kunnath (RadioStudio)}

How about a new year resolution for 2016 and the second half of the decade? How about accelerating the adoption of renewables away purely on economic terms via innovation, especially for fast growing emerging markets like India?

2015 has been a significant year for renewables deployment and policy. The COP21 agreement in Paris has brought together an international consensus and a direction conviction on de-carbonization. In USA, renewables (especially solar & wind) had a lot to cheer in Christmas with the bipartisan extension of the federal tax credits for investment and production (ITC, PTC) with a staged ramp down after a few years. California's public utility commission (PUC) affirmed some of the key provisions (compensating distributed solar) at the retail rate, with some adjustments to fund other rate payer constituents who are unable to adopt solar. While there were a few challenges such as rooftop PV policy in Nevada; Hawaii going towards a wholesale rate (albeit high one) for feed in, the overall policy stance globally going into 2016 is very positive. There has also been a phase out of subsidies in markets like Australia, Germany, Spain where distributed solar has reached higher penetration levels; which raise self-consumption incentives, and hence the early deployment of behind-the-meter energy storage emerging in Australia and Germany. 

Lets turn our attention to emerging markets: I will focus on India. Several of these lessons and technologies will be broadly applicable beyond India. With economic growth rates rivaling or exceeding China for several decades to come, energy is going to be a huge growth market and economic driver in India. The federal government (or central government as its called here) is simultaneously pushing for a huge expansion in baseload coal-fired power, and has set unprecedentedly high renewable energy targets. At 175 GW by 2022 (including 100 GW of solar; 60 GW of wind), this target demands a significant growth from current levels (for example, cumulative solar deployment is ~5GW at the end of 2015). Still at an energy production level, in 2022, 175 GW will represent about 15% of aggregate energy supply from renewables. This implies a huge growth of coal-fired power in parallel. Furthermore, there are significant challenges in the nation's grid to overcome to raise the robustness, reduce technical losses, and absorb high levels of renewables with lower contingency margins compared to global benchmarks. Transmission capacity is also needed to evacuate renewable power (either standalone RE, or bundled with thermal power): green transmission corridors are being rolled out. The government is also driving a massive reform effort (Ujwal Discom Assurance Yojana,  UDAY, in an interesting combination of Hindi & English!) to recapitalize & re-incentivize state-owned or controlled distribution companies (aka "discoms" or "SEBs" in India). 

India is blessed with excellent & widespread solar resource, and low labor costs  ("double the sunshine, half the cost" compared to Japan as observed by Masayoshi Son, Softbank who is making significant investments with Bharti group in India). More specifically, many regions in India get between 5-6 kWh/m^2/day that translates into a energy yield efficiency or capacity factor of 17-23% for solar PV. The engineering, procurement and construction (EPC) costs are also the lowest in the world: 80 US-cents / Wp utility scale or large commercial (compared to almost double that in the USA). Unfortunately India has a higher cost of capital (debt at 10%+, with lower tenors except for the most bankable developers). At utility scale, the government's solar mission reverse auctions have been discovering ever-lower prices (flat Rs. 4.63/kWh or ~7c/kWh  for 20+ years in NTPC AP auctions) and participants. The average power purchase cost (APPC) for many utilities is between Rs. 3-3.50/kWh and private PPAs with thermal power plants (primarily coal) at medium-scale approach Rs. 4/kWh (with escalation for coal input costs over time).  India's wind resources are lower quality (20-25% capacity factor with high seasonality, and site specificity), but expected to be far more attractive at higher hub heights (> 100m) if the economics work out.

So, an interesting question arises: how quickly can solar PPAs fall below Rs. 3 / kWh (i.e. 4.6 c/kWh, below APPC, and below coal)? {Hint: quicker than you would imagine with innovation, even with relatively high cost of capital}

Before we get to that, lets cover a few more points in India's landscape of RE and policy. Renewables are currently a non-dispatchable resource (till energy storage becomes more economical) and impose externalities on the grid due to volatile and seasonal & diurnal (solar) generation profiles. To the first approximation, renewable production is as fickle as the weather, and uncertainty/variability is the enemy of efficient supply chains: this requires continued investment in all forms of flexibility & forecasting/analytics to match demand/supply & grid resources (we call this Cognitive IoT at IBM). Given the limited grid stability margins in India,  the Central Electricity Regulatory Commission (CERC) has rolled out inter-state regulations for grid discipline & integration: specifically, renewable forecasting/scheduling with significant penalties (starting at 10% and upto 30% of PPA) for deviations beyond 15% of scheduled levels. Ancilliary markets and use of energy storage for arbitrage over short time scales are likely to emerge.

All these areas are evolving rapidly, given the central government's focus, the rapidly changing economics of renewables, and by innovative state-level policies. For instance, Karnataka state (where Bangalore is the capital) has attractive open-access solar energy policies, where a developer can install an in-state solar PV plant, and write a power purchase agreement (PPA) contract, not with the utility, but with a private consumer (> 1MW), and the state chips in with waivers of the wheeling, banking and cross-subsidy surcharges. As a result, if you are a credit-worthy private offtaker willing to sign a 10 year solar PPAs (or enter into a group/captive ownership agreement), solar open-access is already cheaper than grid prices. The landscape for roof top solar PV is uneven, with a patchwork of limited net metering policies at states (primarily due to the cash-strapped nature of distribution utilities). Rooftop policies are currently under review in Karnataka, but the prior feed-in tarriff policy of Rs. 9.56/unit led to the Bangalore cricket stadium going solar (though for a variety of reasons, aggregate rooftop PV market penetration has been slow). There is a huge latent market for diesel substitution (during power cuts) as well with rooftop PV (There is an estimated 90GW  of diesel generators in India); and the public sector (esp railways) is a huge sector to target diesel-to-renewable substitution.

In sum, the natural evolution of the solar market in India indicates a strong continued growth in 2016, albeit biased at the ground-mounted side (utility-scale and open-access), and therefore dependent on dynamic policies and execution by the centre and state government procurement, and on the corporate PPA market. The rooftop PV market broadly is nascent, but can grow rapidly (there is a 40 GW solar target as part of the 100 GW target). The off grid market (including solar pumps) is also nascent.

Given this foundation, imagine some of the implications if we were to drive a step reduction in solar LCOE (lets say 20-50%) beyond the natural learning curve cost declines of solar (estimated at 10-20%), and the financial, policy and competitive effects over the next 12-24 months. Here are some possible implications:

    * faster growth of the aggregate solar market in all segments

                  (especially the rooftop, offgrid & open-access markets)

    * the dependence on subsidy can decline in all segments.

    * financial innovation (since lower-tenor loans are easier to get) 

       combined with better project returns can improve accessibility & cost of capital. 

   * broader public awareness, more local jobs will create a more supportive 

       political atmosphere across states and socio-economic classes. 

   * potential implications on the ratio of thermal power vs renewable power to meet the energy needs

   * cheaper daytime power from solar will stimulate markets for energy storage, EVs and 

      day-time manufacturing & enterprises..

Art of the Possible:  Photonic Energy Harvesting

Optics changed the telephony and internet, via high speed fiber optical transmission and cheaper optical networking. Optics will similarly change the course of the solar PV industry especially in markets like India. 

Consider the simple idea: use cheap (commodity) optics to harvest more light (i.e. "cut" light), guide and deliver these additional photons to a solar PV system (i.e. "paste" light). Solar PV systems are  rated to operate consistently at 1000 W/m^2 irradiance, but often see far less light, and have robust packaging for long term warranties (25 years). 

The basic idea of trading off optics for PV is not new: cheap reflectors have been used in concentrated solar PV (CPV) and concentrated solar thermal (CSP) systems for several years. {As an aside, IBM Research has made several contributions to Si-technology, CPV, solar analytics/forecasting, efficient cooling of CPV at a cell level, and High concentration HCPV-T systems.) It is also well known that one of the fundamental ways of improving PV efficiency is through higher light concentration, possibly even with fancier multi-junction cells, if the temperature effects can be managed. 

However, what we are interested is optimizing the asset utilization (i.e. capacity factor or plant load factor) using existing commodity / off-the-shelf PV modules, and view raw cell-level / module-level efficiency improvements as a useful byproduct. Again there is a class of low-concentration PV (LCPV) technologies being researched by several groups, but many have been viewed through the lens of packaging such optics into the PV module, and attempting to couple reflectors with one or more cells. 

Our perspective at IBM Research - India is to "think outside the PV module/box" and view this as a balance-of-system (BOS) design problem, i.e. pull the reflectors out of the module packaging, and manage the "pasting" of harvested illumination using IOT controls & linking these closely with predictive analytics/forecasting etc. This approach allows modules to continue to improve in efficiency and economics; and it can be paired with optical designs & power-electronics flexibly to be customized to a wide range of deployment conditions.  Diffusive optics are cheaper than concentration optics (less precision control, wider variety of commodity/raw materials), can handle both direct and diffuse irradiance (and hence more widely applicable), much lighter than PV  (10x lighter) and can be mounted on a wide variety of low-cost substrates.

The idea of flanking a solar PV module with a mirror or reflector is simple and has been suggested in the past, however the devil is in the (subtle) details to extract performance: different low-cost reflector material/design choices (cost-performance characteristics), dynamic controls to optimize the aggregate irradiance levels, illumination balancing considerations at a PV module & string level; interactions with a variety of module types (c-Si, thin film modules), power electronics (string vs micro-inverters vs DC optimizers), higher production levels even with issues as dust, bird droppings, shadowing (including self-shadowing avoidance), design constraints (space, land area, mobility of reflectors, supplier warranties, UV exposure), integration with analytics/forecasting etc to maximize production and productivity (energy yield for a given level of investment). A large range of auxiliary benefits are also possible, and the system can also be customized for different niche applications (eg: solar pumps, solar CHP, solar chillers/cold storage etc). This is an interesting package of advanced analytics and IoT controls, and has to work reliably for decades at very large scale at a ultra-low opex level.

Our calculations on LCOE indicate economic designs that drive down the LCOE by 30-50% in relatively short order (12-24 months). We assume 15% incremental fixed costs, 12% weighted average cost of capital, and LCOE term of 20 years for <Rs. 3 / kWh, and <10 years for <Rs. 4 / kwh. Importantly, these designs can be customized to be applied to existing solar PV installations (utility, rooftop or offgrid), without technical impact on PV module warranties (business negotiations are another matter). This technology is complementary to the cost reductions in solar PV systems (learning curves in module, inverters etc) . Since reflectors can be made with locally sourceable metals (eg: Aluminium-based designs), and baseline Al-foils are quite cheap (10-100x cheaper than solar PV module/systems), this technology is well suited for a Make-in-India system packaging / EPC revolution. The trick is to manage the cost-performance of the auxiliary elements and remaining in the envelope of contractual terms, maximizing the use of available real-estate (land etc), and handling the performance / maintainence implications (including forecasting/grid integration aspects, automatic cleaning panels / reflectors) carefully at larger scale.  At IBM Research - India, we have worked out these elements and are seeking partners (developers, financial/funding, module makers etc) to transition this technology to market rapidly worldwide, and especially in emerging markets like India, Africa etc.

With deeper integration, future designs will be pre-packaged at the factory or by the EPC, and will admit multiple functions and value streams (eg: hot water, air conditioning, shade, diesel offset, higher pump HP with lower amount of solar kWp, short term solar-driven ancilliary services integrated with storage / power electronics etc).

Summary

Photonic Energy Harvesting offers a complementary pathway to lower LCOE with solar PV systems below Coal LCOE, i.e. sub-Rs. 3/kWh utility scale deployments in India over the next 12-24 months. This kind of cost decline will also have a huge impact in rooftop contexts (especially when handling issues such as shadowing, soiling etc). The cross-over of solar / coal prices will also be an important market signal to drive de-carbonization of the energy mix, and spur early adoption of complementary technologies (energy storage, EVs etc).

Our work at IBM Research - India resurrects an well-known, and old idea (cheap optics to complement expensive PV systems), but we think "out-of-the-module-box" to develop a new avatar in photonic harvesting ("cut-and-paste" light), with a focus on diffusing harvested light rather than concentrating light. The fundamental technical concept is well known for years, but the value is in the customizable technical designs handling the subtle details of packaging/deployment/contracts, economics (driving for the deepest impacts on LCOE), and the integration with business models to open up multiple market segments. A range of technical options that allow superior grid integration via analytics/forecasting (IBM calls this Cognitive IoT), integration & packaging to create higher efficiency system structures, and multi-functional solar+X products make this an attractive direction for the future.

Irrespective of the commercial success/failure of a specific version of the above, or which entity wins in the marketplace, one thing is crystal clear: solar energy costs will come below thermal (coal) energy costs, and (much) faster than consensus thinking. The modularity of these technologies also mean that they can be applied at all scales (utility, rooftop, off-grid) and to a range of applications integrated with solar. These technologies can be applied world-wide in utility-scale and off-grid contexts. The rooftop applicability will be a function of roof type, local regulations etc (eg: flat roofs ideal, but designs available for sloping roofs as well).

ps: If you'd like to explore this technology for your market, or discuss further, please write to me at shivkumar-k@in.ibm.com (official) or shivkuma@gmail.com (personal).

 Acknowledgements:

Immediate collaborators @IBMResearch - India: 

Sukanya Randhawa, Vikas Chandan, Pratyush Kumar, Karthik Visweswaraiah, Rajesh Kunnath (RadioStudio)

Extended teams: Samarth Bharadwaj, Rama C Kota, Amar Azad, Julian De Hoog (Australia), Arun Viswanath (Australia), Sue Ann Chen (Australia), Vijay Arya, Ashish Verma, Babitha Ramesh, Saravanan Jagadeesan.

Executives: Ramesh Gopinath, Chandu Visweswaraiah, Robert Morris, Zachary Lemnios.