As described in our previous dialog, a number of new market factors are stressing utilities’ ability to deliver 3 x 9s reliability.
These factors fall into four categories: (1) new or expanded energy policies and regulations, (2) deployment of SG applications absent reliability-enhancing SG controls, (3) imperfect coordination between electricity market-clearing processes and the physical control processes of the power system, and (4) aging power system infrastructure.
Evolving Energy Policies and Regulations Have the Potential to Negatively Affect Reliability
Utilities dispatch power generation based on a net load forecast, where net load equals the native customer load minus any power generated by (1) a self-optimizing individual customer (e.g., distributed generation or energy storage discharge), (2) an aggregated self-optimizing set of customers, or (3) a micro-grid.
RPS energy policies as well as regulatory policies encouraging DG, EVs, distributed storage, CHP, and micro-grids are having increasingly significant effects on the shape of the net load and on the first derivative of the shape. For example, Mike Niggli, CEO, SDG&E, speaking in Distributech 2013’s plenary session, referred to expected load ramp rates in March 2020 of 4,500 MW down in two hours and 12,500 MW up in two hours, on a 25,000 MW system.
In most cases, the utility does not have visibility into customers’ distributed generation decisions ahead of time. The challenge for the utility is to maintain its target level of service reliability despite the uncertainty associated with the ensuing net load.
To a certain extent, short-term volatility in the net load caused by intermittent generation (distributed PV) may threaten system stability, especially if aggregated. Some utilities have established rules of thumb for the maximum percentage of PV they will allow on a feeder, e.g., 15%. However, it appears that these rules of thumb/heuristics are overly conservative. One private study simulating a typical distribution system found that its feeders, even in low load situations, could tolerate PV capacity of more than 50% of the load when appropriate (and not too complicated) control equipment is put in place.
To decrease the uncertainty in the net load forecast, and to access additional existing capacity next to the load center that can help maintain reliability in tight supply situations, some utilities offer a “virtual power plant (VPP)” program to their customers. For example, ConEd, PGE, CPS Energy/San Antonio, Duke’s Microgrid Program, AEP, and Europe’s FENIX program offer VPP programs of different types.
In some of these VPP programs, the utility interconnects, maintains, and operates the customer-owned generation/demand reduction applications as a bundle of dispatchable capacity, in return for which the utility provides the customer with certain tariff concessions.
Jurisdictions offering dynamic pricing, e.g., TOU, CPP, and RTP, also create uncertainty in the load forecast. Automated customer price responses can produce large, rapid, swings in the net load. If the consumer’s price response is not automated, i.e., not “smart”, the net load forecast uncertainty can likely be reduced over time based on increasingly accurate (“learned”) estimates of the price elasticity of customer segments -- it helps that price responses will likely be diversified across the service area.
To incentivize an acceptable level of service reliability, state regulators in over 50% of states have mandated penalties for SAIDI or CAIDI performances above a predetermined acceptable range, or have instituted service quality mandates with quantitative metrics. The penalties can be costly -- they provide a strong incentive for utilities to install equipment that improves reliability.
Naturally, these equipment costs are subsequently reflected in customers’ bills. However, the solutions simultaneously improve utility asset utilization and can even prolong the lifetime of some utility assets.
Somewhat Surprisingly, Initial Deployment of SG Applications Can Have a Negative Impact on Reliability
While SG applications can help enhance reliability through smart sensors and increased automation, it appears that the initial SG applications could negatively impact system reliability before subsequent D.A. applications provide ameliorating automation, i.e., SG can be first a sword against, and later a shield for, reliability.
Here we will address the negative impacts and follow-up below with some business opportunities for SG applications that mitigate these negative effects on reliability.
The essential attribute of SG deployments is automation -- automation to improve asset utilization and efficiency in the generation, delivery and end-use of electricity. This automation is in the form of machine-to-machine (M2M) interfaces or hybrid M2M/human interfaces which are pre-programmed (for the most part) to (a) react to changes in electricity prices, (b) optimize the use of resources, and (c) maintain service reliability.
In the context of the SG, M2M interfaces are incorporated, for example, in smart thermostats, pool pumps, air-conditioners, water heaters, washing machines, distributed generation, integrated self-optimizing customer control systems, and building automation systems. Automated demand response is an M2M-controlled system. These M2M interfaces can (a) react rapidly to changes in electricity pricing, time of day, or weather, (b) mitigate incipient reliability problems, and (c) optimize the use of assets in near-real-time. M2M systems sometimes manage aggregate loads from different facilities, magnifying the amount of uncertain load under the same program of automated control.
M2M systems react very rapidly when conditions deviate from their set points. There is potential for the changes in electricity demand (a.k.a. dynamic demand) created by M2M automation to out-speed the human operator’s ability to dispatch power to cover these rapid changes in load -- potentially causing reliability or stability problems in the delivery system.
The conclusions from a fascinating paper simulating the potential for smart grid-caused instabilities resulting from interactions between market-based (i.e., price) feedback loops and physical control systems are summarized below.
In contrast, SG applications in the form of more sophisticated automated controls can be used to mitigate the potential for reduced system reliability, e.g., by optimizing the operation of distributed storage to provide ancillary services, or by building randomness or time-delays into M2M control algorithms.
If we believe that maintaining 3 x 9s system reliability could be negatively affected as a result of the deployment of salutary SG applications, it is logical to believe that there will be new business opportunities associated with providing solutions to these potential service reliability problems. The solutions will likely be in demand by utilities as PUCs continue to tighten mandated minimum levels of reliability across the U.S., and by end-use customers who seek to address service reliability problems from behind the meter.
The Impact of Market Behavior on Reliability
Rick Geiger, Executive Director, Utilities and Smart Grid, Cisco Systems Inc., refers to the possibility of a “flash crash” as we engineer a "price-responsive, transactive grid". He bases this possibility on the penetration of SG active control systems into a system that currently has more electro-mechanical control devices than intelligent ones. A broad, highly price-responsive market could automatically trigger hundreds of MWs of demand to drop off the grid successively at pre-programmed increasing price tiers. This would occur outside of the grid’s physical control system, i.e., powered by financial transactions, and could constitute high-gain positive feedback loop. Rick calls for a “new control model” to manage this risk.
Could this “flash-crash” happen? The answer is yes -- a recent fascinating paper by Masiello et al. shows that instability is a distinct possibility under some quite realistic assumptions. The study used a system dynamics model to simulate three feedback loops in the power system: (1) the demand response of the customer to a price increase with a time lag, (2) the supply-side response to the change in demand, with a time lag, and (3) the market-clearing mechanism activating to re-balance supply and demand, with a time lag.
They identified a number of scenarios where instability propagates as a result of the interaction of these three feedback loops. For example, if the time-lag associated with the response of demand to price changes exceeds the frequency of posted changes in real-time pricing, instabilities can occur.
It appears that instability is more likely to begin near the time of the peak load because it requires a relatively high price level to trigger instabilities. The instabilities may, or may not, persist, depending on how the price elasticity changes after that time. Errors in estimating the price elasticity of demand within the market clearing process can also lead to instability.
The above correlated behavior between the feedback loops seems intuitively logical.
Another interesting, though not unexpected, finding by the authors was that relatively small amounts of aggregated self-optimizing customers’ loads can also cause price instability. This is because M2M pre-programmed responses of, for example, smart appliances, can out-speed feasible changes in load-following power generation. As self-optimizing customers and micro-grids increasingly penetrate the end-use market, it will be important for the operators of the power system to have better information about, and visibility into, customers' behavior, if they are to maintain traditional levels of service reliability.
We hear again and again that a substantial proportion of power system equipment in the U.S. is more than 40 years old, i.e., past its design life. There is justified concern that this situation will lead to deteriorating service reliability, as measured for example by SAIFI (frequency of failure occurrence) and SAIDI (duration of outages) metrics.
Two founding partners of Quanta Technology LLC, H. Lee Willis and Dr. Richard E. Brown have pioneered work on aging power infrastructure management. In their recent must-read, excellent exposition, they state: "The industry will ultimately have no choice but to either accept worsening reliability or somewhat higher costs.”
No doubt, aging infrastructure is a growing reliability issue. As power system infrastructure equipment continues to deteriorate, the longer we wait to arrest this decline the more costly and longer it will take to bring it back to acceptable reliability performance, i.e., where failure rates are constant, and at an acceptably rate of occurrence. The Quanta authors above call this level of performance the equipment “sustainable point”: an operating status at which "the infrastructure no longer deteriorates and performance no longer worsens each year".
But resolving the aging infrastructure issue does not necessarily involve replacing large swaths of infrastructure. The challenge is to identify which equipment is going to fail first in order to “surgically” replace the worst equipment, while avoiding the replacement of equipment that can continue to function well, even though aged past its design life; and then, maintaining this relatively low percentage gradual rate of replacements ad infinitum, mirroring the aging rate of the infrastructure. Mr. Willis and Dr. Brown make a very cogent case for this approach.
In the SG environment, existing data bases (e.g., AMI, SCADA, GIS, OMS, CIS, data historians) in combination with advanced power flow simulation models and cost-benefit analyses can be used to prioritize the replacement of equipment in electric distribution systems. Gridiant Corp. is an example of a software company that provides this type of capability—it optimizes equipment replacement programs based on economics using a high-fidelity two-way power flow model.
The following section presents some SG applications that may create reliability-related business opportunities.
Smart Grid Applications Addressing Reliability Issues
We can use an in vogue taxonomy to categorize some of the business opportunities associated with the service reliability issues presented above: situation awareness, situation intelligence/analytics, and situation control; and we’ll add our own category: new business concepts.
Situation Awareness Business Opportunities
Measuring the status of nodes, feeders, circuits, transformers, switches, and substations in the distribution system
- Smart, communicating sensors, including micro-synchrophasors
Situation Intelligence/Analytics Business Opportunities
Applying analytics to selected information from existing data bases, and data feeds from smart sensors (referred to as "big data" these days) -- distribution system software and associated algorithms to:
- Compute real-time state estimates
- Prioritize and justify capital expenditures on reliability
- Develop preventative maintenance programs (“surgical” protection versus “blunt instrument” protection deployments)
- Develop short-term load forecasts
- Optimize operations of self-optimizing end-use customers, micro-grids, virtual power plants
Situation Control Business Opportunities
Development of smart grid controls:
- Smart switches and power management devices
- Smart inverters
- Advanced control algorithms for real-time and near real-time applications
- More efficient algorithms for solving ultra-large system problems
- Rapid power restoration
- Management of line losses, voltage levels, volt-amp reactive (var) power
- Management of major coincident loads
New Business Concepts
- Utility dispatch of customer-owned distributed generation, i.e., virtual power plant programs (VPP)
- New products for power markets that goe beyond the current set of ancillary services, e.g., flexibility supply, reliability supply, inertia supply
- Smart Demand-Side Management (SDSM) – dynamic load management (the equivalent to DSM 2.0)
- Integrated Distribution Automation Infrastructure (DAI), akin to AMI
- APIs that connect and coordinate wholesale markets with retail markets
As always, comments welcome and appreciated in the comment box below.