Open access peer-reviewed chapter

# Risk Mitigation in Energy Efficiency Retrofit Projects Using Automated Performance Control

By Job Taminiau, John Byrne, Daniel Sanchez Carretero, Soojin Shin and Jing Xu

Submitted: May 29th 2019Reviewed: September 2nd 2019Published: March 10th 2021

DOI: 10.5772/intechopen.89476

## Abstract

Performance gap concerns limit investment in the building energy efficiency retrofit market. In particular, the ability of projects to deliver on promised energy savings is commonly drawn into question. Performance risk mitigation mainly occurs through energy saving performance guarantees. Contractual stipulations arrange the conditions of the guarantee, and ceteris paribus, a higher energy saving guarantee should reduce project performance risk. Therefore, methods that yield a higher energy saving guarantee could help accelerate the market. We review the ability of “smart,” automated, and connected technologies to: (a) intelligently monitor and control the performance of energy-consuming devices to reduce performance variations, (b) provide additional degrees of control over the project’s performance, and, by doing so, (c) motivate the energy services company (ESCO) to raise the energy saving guarantee. Our analysis finds that use of such automated performance control could significantly raise the energy saving guarantee, making projects more likely to succeed.

### Keywords

• derisking energy investments
• energy efficiency
• performance gap
• energy savings guarantee
• building controls
• Monte Carlo analysis
• monitoring and verification

## 1. Introduction

### 3.1 Benefits of automated controls

Implementation of automated building controls could help prevent project under-performance. The potential for this technology option is substantial. For example, an assessment by the U.S. Energy Information Administration (EIA) documented in 2012 that over 85% of commercial buildings in the United States have inadequate control infrastructure in place ([53], as quoted by [54]). In addition, it has been broadly established that advanced control measures can improve performance and save 10–30% of energy consumption [20, 54, 55, 56, 57, 58]. For lighting, for instance, a combination of improved lighting devices and controls can reduce commercial lighting energy use by 81% [59]. A meta-analysis looking at the savings generated by lighting controls in commercial buildings by isolating the control function found savings ranging from 28 to 40% with combined operation of sophisticated controls achieving higher saving rates [60].

Whole-building energy management systems integrate a variety of end-uses (including services beyond energy such as security). A survey of zero net-energy buildings that use building controls found that 91% of the commercial buildings surveyed in North America relied on control systems that integrate multiple end-uses with 67% using a fully integrated controls architecture capable of controlling all end-uses centrally [61]. Many of these systems do often still rely on the occupant for some part of the successful operation of the controls: 74% of the buildings surveyed have integrated controls system sequences that are not fully responsible for driving performance, relying instead on the occupant [61].

Relative to the potential, significant under-adoption of the technology suites can be observed and this is often attributed to the high cost associated with whole-building applications [27]. The suite of technologies is typically deployed as software as a service (so-called SaaS) offerings, delivering capabilities on a subscription-type basis [27]. In other words, up-front expenditures for items such as licensing and system configuration are accompanied by recurring subscription fees which spread out the cost of the entire system over its lifetime. Nevertheless, up-front cost estimates range from $10 to$3400 per point with most in the $100 to$500 per point range [27]. In addition, the recurring costs range from $5 to$3100 per point [27]. Put together, 5-year ownership estimates ranged from $140 to$16,000 per point [27]. A point is a single datum that is trended, stored, and available for normalization and data analysis across use cases and comprehensive, whole-building systems can have thousands of points. For example, a use cases overview of a major controls company shows how a project involving three federal office buildings contained 18,000 points [62]. Therefore, at the median 5-year ownership costs found by Ref. [27] of $1800 per point, a fully integrated energy management system could cost as much as$32 million for the three federal office buildings.

Strategic design and selection of automated control technologies at the end-use level might overcome this barrier. Versions of partial integration deployment strategies can be observed in the market: the survey of zero net-energy buildings found that partial integration of end-uses occurred in 24% of the buildings while 9% had no whole-building controls architecture at all but, instead, used controls only at the end-use level [61]. At this level of operation, there is an expectation of significant cost reduction to the point where control technology cost can be brought down from an estimated $150–$300 per node to $1–$10 per node using low-cost, self-operated, and wirelessly connected end-use level devices [63]. Lower costs opens the door for automated controls to fulfill performance control functions for key ECMs.

Coordinated implementation of end-use level automation could enable projects to reap additional benefits:

• Operational and engineering riskreductions include time efficiency, improved accuracy, and possibilities for standardization and certification. For instance, automated performance control accelerates whole-building assessment from a typical 4 days to 1 day and reduces time needed for custom engineering calculations from 6 days to 1 day [64]. Automated analysis yields actionable data within the first project month [57]. Analysis of 537 projects further shows that industry standard predictive accuracy can be achieved with only 6 months of training data [65, 66]. When assessed as part of a portfolio of buildings, predictive accuracy improves further leading to the conclusion that these models are “compellingly accurate” [64]. Automated data analysis enabled by automated control enables attribution of consumption pattern variation, standardization and certification—a key need of the sector to develop investor-ready program design [7].

• Monitoring and verification riskreductions include portfolio level analysis, benchmarking, improved sampling, and fast anomaly and fault detection. Real-time and high-resolution automated control of performance makes even small or portfolios of projects capable of processing “big data”. Further, scalability and precision allows larger sample sizes, retrieving feedback on the performance of diverse aspects of the retrofit project. Real-time data collection and control enables faster anomaly or fault detection and interface options such as online dashboards empower clients and ESCOs to mitigate under-performance.

• Economic riskreduction benefits from automated control include real-time utility tariff and energy consumption analysis to validate utility bills through, among others, (a) continuous monitoring and management of peak load consumption; (b) streamlining of utility-related processes to, for example, minimize personnel requirements; and (c) identification of metering or billing errors by automatically crosschecking consumption patterns with utility bills.

• Financial riskreduction is achieved through uncertainty mitigation and improved project finance-ability. Strategic use of automated control can deliver investor-ready program design and enhance the energy savings guarantee. Data generated by automated control can improve project finance-ability as it supports, among others, (a) accurate savings estimates, (b) risk management of operational and performance uncertainty, and (c) quick remediation of potential energy saving shortfalls.

### 3.2 Modeling the contribution of automated controls in the energy savings guarantee setting process for building retrofit projects

The level of savings that an ESCO will guarantee is principally influenced by: (a) the project’s ability to confidently deliver savings and (b) the risk tolerance of the ESCO. A simplified interaction dynamic between project host and ESCO is provided in Figure 2. For a hypothetical project, Figure 2 shows that the project’s savings can exceed a low guarantee but will likely fall short when a (very) high guarantee is used. Under a guaranteed savings structure, savings above the guarantee are awarded to the project host while savings that fall short of the guarantee negatively impact the ESCO. This is illustrated in Figure 2 by the green and red areas, respectively. From the perspective of the project host, savings that exceed the guarantee are welcome but, critically, these savings are not guaranteed and, therefore, are not available to underwrite the initial investment. However, savings that fail to reach the level of the guarantee prompt the project host to argue for compensation that could be disputed by the ESCO. This can be an arduous process. The project host, overall, is interested in a high guarantee but concerned about disputes with the ESCO [11, 12].

Surplus savings above the guarantee occur when the project over-performed relative to the guarantee. In this case, the ESCO is not at risk of claims for compensation. While this sounds appealing, it also means that the project bid by the ESCO could have been more competitive. Insufficient savings to cover the guarantee lower the overall return on the project and could even represent a net-loss for the ESCO. From the perspective of the ESCO, this should be avoided whenever possible. As illustrated in Figure 2, a hypothetical range of possible guarantee values that are acceptable to the ESCO can be identified. In general, the ESCO is incentivized to place the guarantee below expected savings but is hesitant to place it too low.

So far, we have established that energy efficiency savings performance can be uncertain. One way to consider this uncertainty is to reflect on energy savings performance as a stochastic distribution of possible savings. A broad distribution of energy savings i.e. a higher probability for adverse circumstances as listed in Table 1 presents a higher risk for the ESCO that performance levels will be below the guarantee (e.g. [67]) and, ceteris paribus, is accompanied by a lower guarantee. To estimate these considerations, we follow the model proposed by Deng et al. [41]. This approach calculates, from the perspective of the ESCO, the annual and total profit for a series of possible guarantees against a Monte Carlo analysis-derived savings profile. We use the approach to calculate the guarantee level where, for a given risk tolerance, the ESCO will is unlikely to experience losses due to insufficient savings and resulting claims for compensation by the project host. In other words, consider a Monte Carlo simulation of a project’s performance that results in a stochastic distribution of possible savings. In this case, a 95% risk tolerance would result in a guarantee level where 95% of all the simulated outcomes deliver savings sufficient to cover the guarantee in each year of the project lifetime. Then, the maximum guarantee within the ESCO’s risk tolerance is selected as a probable guarantee for the project.

The steps of the analysis are to, first, derive possible performance profiles for pre-retrofit, post-retrofit without controls, and post-retrofit with controls scenarios for each year of the hypothetical project. This step produces three distributions of performance that approximate normal distributions. The contribution of building controls, here, is to substantially narrow the distribution of post-retrofit performance, leading to more secure savings profiles. In addition, controls improve actual building operations, leading to a savings profile with a higher overall average savings. The second step is to take the probabilistic savings profile and compare it against many possible guarantee values to identify the moments where the savings fall short. From the perspective of the ESCO, any moments where the savings exceed the guarantee are set to zero (these savings are awarded to the project host). Finally, within the stated risk tolerance of 95%, the maximum guarantee where savings are sufficient to cover the guarantee is calculated.

### 3.3 Software stack and data inputs

The primary software element is the U.S. Department of Energy‘s (DOE) Energy Plus software: a leading building energy simulation tool in the energy efficiency industry [68, 69, 70]. Advantages of Energy Plus include first-principles, text input–output work-flow that can be automated [71] and availability of benchmark building model databases (16 building types across 16 locations and three construction periods) [72, 73]. Within Energy Plus, we made use of DOEs prototypical commercial building models that describe typical building layout, geometry, energy consumption, etc. for buildings in the Delaware region constructed before 1980 [73, 74]. In particular, we model the performance of the “large office” building benchmark. The energy performance of the large office benchmark building was simulated using Energy Plus version 8.6.0. This DOE benchmark building reflects a possible building operated by the public sector, the dominant user currently of energy savings guarantee projects.

The large office benchmark building is a 46,320 square meter, 12-story office building (including basement) with total annual baseline consumption of 26,358 GJ of electricity and 7266 GJ of natural gas to fulfill its end-use functions or 725.9 MJ/m2. Notably, over half of the buildings energy consumption serves interior lighting (9422.03 GJ or 28.1%) or interior equipment (8384 GJ or 25.2%). Heating is third most responsible for annual energy consumption (7265 GJ or 21.6%).

Possible ECMs were identified using research results from Lawrence Berkeley National Laboratory (LBNL), specifically the Commercial Building Energy Saver (CBES) project (http://cbes.lbl.gov/ and Refs. [75, 76, 77]). This ECM selection was further supported by data from the Building Component Library and several articles using a similar methodological approach [11, 36, 43, 71]. Finally, our research team had access to guaranteed energy savings agreements (GESAs) provided by ESCOs for other projects in Delaware and across the United States. Data from these GESAs was used to complete ECM profile selection by looking at buildings in those projects that share similarity with the benchmark building. Critically, based on a review of existing control literature, the selected ECMs listed in Table 2 can be accompanied by automated controls.

ECM & nameEnergy plus parameter (Unit)ECM costs
1. LED lighting upgradeLighting load (W/m2)$0.63/m2 [78] b 2. Appliance upgradePlug load (W/m2)$5.29/m2 [79, 80]
3. Thermostat set-point updateSet-point in Celsius (C)$49.10/thermostat [80] 4. Chiller replacementReference COP (fraction) a$439.48/ton [80]
5. Boiler replacementNominal thermal efficiency (fraction)$34.96/MBH [80] 6. Install high-efficiency fansFan total efficiency (fraction)$0.176–$0.390/cfm [80] 7. Water heater replacementHeater thermal efficiency$20.82/gallon [80]

### Table 2.

Selected ECMs and key parameters.

Coefficient of performance.

Typical retail prices for LED packages purchased in quantities of 1000 from major commercial distributors. Using price point estimates for 2020 for cool white LED packages at 218 lumens per Watt.

Parametric evaluation of the building models was conducted using jEPlus software (version 1.7.2), an open-source parametric analysis tool specifically designed for Energy Plus [71] that provides flexible and structural analysis opportunities and smooth operations [81]. The tool has been used in similar investigations to determine sensitivity or optimize energy systems [82, 83]. This set-up enables Monte Carlo analysis for risk estimation and management of, among others, renewable energy projects, system planning, or system optimization [84, 85] and for energy efficiency projects in general and monitoring and verification efforts specifically [36, 44, 82]. Latin Hypercube Sampling (LHS) was used to run 10,000 simulations per jEPlus model run. LHS is a powerful tool that enables efficient stratification across the uncertain performance range [86]. Parametric evaluation was conducted on Amazon Web Services (AWS) architecture. The data inputs used for the parametric evaluation are provided in Table 3. Controls are assumed to be able to reduce performance variation by 90%.

ECMDistributionInputPre-retrofitPost-retrofitSource
1NormalValue:16.146.46[82]
σ:0.5650.226
2NormalValue:10.768.07
σ:4.5493.412[82]
3Triangular aHeating[82]
Value 1:2120
Value 2:15.614.6
Min/Max:±6.52%±6.52%
Cooling
Value 1:2425
Value 2:26.727.7
Min/Max:±6.52%±6.52%
NormalValue:5.116.2[82]
σ:0.0240.029
NormalValue:0.760.95[44]
σ:0.0110.014
NormalValue:Various0.65σ = 5%
σ:0.0500.033
NormalValue:0.80.95[44]
σ:0.0120.014

### Table 3.

Pre- and post-retrofit performance variation inputs for the large office.

Two thermostat threshold set-points for heating and two threshold set-points for cooling are included in the model.

## Acknowledgments

This research was supported by the Delaware General Assembly.

## Conflict of interest

The authors declare no conflict of interest.

## Notes

Supplementary information, data, and models are available at http://freefutures.org/publications/.

## Notes

• Other contractual agreement forms, such as first out or chauffage, are also available but are not evaluated here.

chapter PDF
Citations in RIS format
Citations in bibtex format

## More

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License, which permits use, distribution and reproduction for non-commercial purposes, provided the original is properly cited.

## How to cite and reference

### Cite this chapter Copy to clipboard

Job Taminiau, John Byrne, Daniel Sanchez Carretero, Soojin Shin and Jing Xu (March 10th 2021). Risk Mitigation in Energy Efficiency Retrofit Projects Using Automated Performance Control, Sustainable Energy Investment - Technical, Market and Policy Innovations to Address Risk, Joseph Nyangon and John Byrne, IntechOpen, DOI: 10.5772/intechopen.89476. Available from:

### chapter statistics

1Crossref citations

Next chapter

By Ryan M. Yonk