*4.2. Implementation and Validation of the Artifact*

DSR guidelines [43] directly stated that upon creation, the artifact needs to be evaluated (G3) and tellingly reported to interested parties (G7). In our case, both of those scientific rigor-oriented steps were effectuated jointly by the research teams and business partners as a part of the project aimed at enhancing EE within the network of facilities owned/franchised by the Polish subsidiary of a global player representing the petroleum industry. Royal Dutch Shell PLC, headquartered in the Netherlands, is included on a Global 2000 List of the World's Largest Public Companies, with \$394 billion total assets and over \$311 billion of revenue [44]. Its retail arm by the end of 2019 covered approx. 45 thousand facilities that operated based on different business models in close to 80 countries and handled over 30 million customer visits on a daily basis [45].

The pilot works implementation process unveiled two deviations from the original design of the artifact, i.e., the number of areas for EEM covered by pilot works and the length of the pilot itself. While the latter might be classified as minor, the former deserved special consideration. Pilot works

were launched by selecting a pool of facilities (in this case: petrol stations) from among a total of over 400 facilities located in Poland. The methodology provided for designating two groups of facilities—with the first group being measured only at the main power input, and the second configured with multiple meters. Electrical circuits measured by the more sophisticated configuration included interior lighting sources of the station, air conditioning, and exterior lighting sources of a station (roofing, driveways, illuminated advertising, etc.). All the determinants necessary to initiate analytics (locations, weather conditions, volumes, schedules of operation, relevant contractual details between the customer and all energy suppliers involved) were provided, integrated, and double-checked.

Installing respective meters paved the way for collecting data on electric power consumption. Upon considering both the needs of the participating organization and the technical constraints, it was decided to send aggregated data packages from individual facilities every 15 min. Electric power consumption was plotted as hourly bars and supported by the BI system in place. It related power consumption to the aforementioned determinants. Simultaneously, the collection of additional data to support these activities was launched. In line with the construction of the artifact, energy audits were conducted as a part of pilot works, and technical inspections of electrical devices, thermo-vision, and building/woodwork audits constituted the primary sources of additional data. Values and informational content, upon being subjected to analysis, influenced the EEM of each petrol station involved.

Of the five areas of activity for EEM that the initial concept envisaged, specific actions to be completed were listed within areas 1, 4, and 5 (see Figure 1). Areas 2 and 3 were excluded from the pilot works. Activities in the areas undertaken for implementation were carried out until the end of the pilot's timeframe. The work was concluded with several recommendations, including guidelines for staff behavior at stations, benchmarking the existing electric power tariffs against optimal ones, selection of new tariffs for each facility, and proposing reductions in contracted power levels. It was empirically confirmed that the actions within these three areas resulted in a net reduction in power costs at the designated petrol stations without compromising business continuity.

Ultimately, the pilot EEM implementation lasted from 1 January 2019 to the end of May. Project inception required 2 weeks and involved preparing measurement equipment, assembling installation teams, training on safety at work at petrol stations as well as scheduling telemetry installation. The actual deployment of telemetry in all five facilities totaled one month. Beginning in mid-February, every telemetry system started generating measurement data and forwarding it to the data processing IT solutions. While measurements were kept live until the end of the pilot works, at the end of February, early feedback and recommendations were determined. Between the beginning of March and the end of May, recommendations were introduced in all facilities, and the results were observed. Upon the pilot project completion, concluding workshops were held.

As energy services tend to be settled under performance-based contracts [27], the pilot served as a data source for elaborating pessimistic, realistic, and optimistic scenarios for increasing EE during the course of the actual contract. The pilot showed a statistically significant potential to reduce the nominal energy consumption. The specificity of petrol stations' operations (24/7 mode, excluding the short period where the customer-side systems settle daily transactions) contributed to power consumption not being reduced by late evenings and during night times. On the contrary, consumption is statistically lowest during (late) mornings and mid-day. This is when few light sources are used, and the Central European climate implies less intensive use of heating systems. Regardless of the above, the selection of energy tariffs, which were rational when signing contracts with energy suppliers, became less favorable over time. Tariff systems tended to become increasingly complicated, and the unbundled energy distribution market in Poland makes identifying optimal solutions on one's own, time-consuming. In this case, having an up-to-date tariff database that is for the needs of the entire customer portfolio significantly facilitates optimization processes. Whereas tariffs individually tailored to the profile of each facility where possible, show that time periods in which energy prices are higher than the base

tariff (see Figure 2, the median case with a linear tariff), are more than compensated for over the rest of the day.

**Figure 2.** Differences in energy consumption and tariff adjustment.

Given the time necessary for reflection, the participating organization committed to up-scale their EE-oriented efforts to the entire network of Polish facilities by signing a 3-year contract at the end of June 2019. The pilot's pessimistic scenario for increasing EE was adopted as the benchmark for the contract. In its first year, which also covered the 5-month-long telemetry installation phase, the provider of the energy services took upon oneself achieving cumulative savings of 2.25% on one-third of the base amount. Throughout the second year of the contract, a significantly higher level of EE improvement was set: no less than 10% on two-thirds of the base amount. Such a scale of commitment was in line with the long-term estimates formulated in the post-pilot reports and the adjustments for unsupported areas of activity (see Table 2). The calculation took into account both the saturation of the telemetry system with data and the development and implementation of specific policies, as well as getting to know the facilities, personnel, and internal processes of the petrol company. Pilot-based projection indicated the need for a conservative approach in the final year (cumulative savings of 14% of the total base amount) due to reaching expected limits regarding opportunities for reducing electricity consumption without affecting the continuity of business.

**Table 2.** Post-pilot estimates regarding the scale of savings in the long run.


#### **5. Discussion**

The study demonstrated that EEM is, on one hand, a long-term process that requires continuous monitoring and improvement as well as adapting to constant changes both in the business environment and within the organization. While the necessity to take advantage of enhanced procedures and adapt business processes were explored before [35], lessons learned from this study are somewhat polemic with previous conclusions that companies are generally uninterested in ventures lasting beyond 3 years [17]. The recommended best practice in this respect is to stage the target contract so that periodical milestones/review points and success fees were considered (Table 3). In this very case, the progress is being monitored at weekly and monthly intervals. On top of that, once a year an official presentation of the savings recorded and details of the methods to achieve them is given to a wider audience of customer-side professionals, as indicated in the contract.

On the other hand, such a contract proves parameterizable to a large extent. The study showed that EEM implementation features selecting a set of options that do not exhaust all the possibilities but are favorable in a given horizon and budget/investment capabilities. In this case, three EE-improvement areas (human behavior, tariffs, capacity management) indeed required an investment in a telemetry system as well as developing and implementing an EEP. However, compared to the other preselected areas, it involved no additional costs related to upgrading/replacing electrical systems or conducting a general overhaul of a facility or its components (such as woodwork). Such fundamental ventures depend on the firmness of financial commitment from the investor and the lack thereof put the capability of meeting the conditions for success fee at risk. Hence, the recommended best practice in this regard is to scale down the list of potential EE improvement areas within the first 30% of the contract.

First-year performance indicators exceeded both the contract reference values and the average savings of approx. 5.0% reported by Fresner et al. based on the analysis of 280 EE-targeted projects across seven different countries [17]. It should be pointed out that the adoption of the pilot's pessimistic scenario did not increase the accuracy of forecasting the actual performance of the contract, which might be assumed based on the study by Fowlie et al. [46]. Similarly, actual energy savings in Zivin and Novan's analysis were off by 21% [47]. Nevertheless, the energy service implemented following the application of the methodology was beyond the scope of the energy audits covered by those analyses. Moreover, small/medium-sized enterprises have a natural tendency to avoid the more intrusive measures that a corporate body can easily afford, and whether the COVID-19 pandemic has had a noticeable impact on the results remains to be seen. Adopting realistic scenarios as benchmarks for similar contracts in our opinion requires pinpointing methods for calculating savings generated, which will take into account factors such as volatility of electric power prices, weather anomalies (atypically warm winters or hot summers), as well as fluctuations in equipment lists of relevant petrol stations. The latter involves, in particular, extra electrical hardware such as HVACR devices or (super) chargers for electric cars.

There is a need to highlight the strong feedback of the practitioners that the success of the project and assuring the efficacy of developed measures requires the involvement of employees from various levels, as well as adapting some business processes. This is in line with Johansson and Thollander's contribution, who listed top-management support as a leading success factor regarding EEM practices [16].


**Table 3.** Key activities within the EEM framework.
