# From one asset to a network Level two · L2-NETWORK · 12.09.2026 Transferability, asset differences and the cost of support at scale. ## Starting asset Teaching portfolio N-01: six A assets at 100 MWh/year with a 20% effect, three B assets at 200 MWh with a 15% effect, and one C asset at 1,000 MWh with a 5% effect. Implementation costs per asset are €10,000, €20,000 and €100,000 respectively. Annual support is assumed to cost €600 per asset. The scenario tests scaling methods. ## Outcome Defend a rollout across different assets and establish transfer limits. ## 1. Describe the transferable solution After the first project, identify the physical mechanism, required data, applicable operating conditions and support requirements. Success may depend on a particular pre-existing defect, staff behaviour or contract feature. Include those conditions in the solution record. Transfer requires stable principles and parameters that must be recalibrated. A solution version includes model, training data, settings, metering design and stop conditions. Screen each new asset for compatibility. The pilot result supports a forecast with explicit limitations. ## 2. Segment the portfolio by meaningful differences Group assets by equipment, climate, schedules, temperature requirements, metering quality and available actions. Building type or brand alone may explain little of the outcome variation. Each group needs a representative test asset. The large C asset in N-01 has a low percentage reduction but a substantial absolute effect. Its rollout priority also depends on implementation and support cost. Keep poor-performing assets in the analysis with their causes. Unexplained exclusion of difficult cases biases the network forecast. ## 3. Use the correct aggregation weight N-01 baseline demand is 6 × 100 + 3 × 200 + 1 × 1,000 = 2,200 MWh. Group savings are 120, 90 and 50 MWh, totalling 260 MWh or 11.82% of baseline energy. The mean percentage across ten assets is 17%. It describes the average asset percentage, not the share of energy saved by the whole network. Power, money and reliability need their own aggregation units. Aggregate peaks with coincidence and network constraints taken into account. ```text Portfolio_rate = Σ (baseline_i × saving_rate_i) / Σ baseline_i ``` ## 4. Test a held-out group Use some assets for tuning and hold others out for independent testing. Compare forecast and actual performance by asset type, explaining major deviations. With few assets, show individual outcomes and the covered operating range. Zero failures in twenty trials does not establish failure-free operation. For independent identically distributed trials, the exact one-sided 95% upper failure-probability bound after zero failures is 1 − 0.05^(1/n). At n = 20 it is approximately 13.91%. Shared causes and asset differences violate that simple model and need separate analysis. ## 5. Plan deployment waves Each wave specifies assets, solution version, acceptance tests, observation period and a condition for pausing expansion. The next wave receives corrections and an updated forecast. Changing equipment, model and contract simultaneously complicates causal analysis. Plan per-asset rollback and containment of shared errors. Loss of communication triggers a defined local mode. A central agent sees data status and available flexibility while retaining owner constraints. Assess network value after checking node behaviour and interactions. ## 6. Include delivery and support Budget surveying, equipment, configuration, connectivity, metering, visits, updates and replacement. N-01 implementation costs €220,000. At an illustrative €0.18/kWh, gross annual avoided-energy value is €46,800. Supporting ten assets leaves €40,800 before other costs. Simple payback of about 5.39 years excludes discounting and changing cash flows. Portfolio growth increases support workload. The laboratory assumes 16 commissioning hours and 24 annual support hours per asset. These are teaching allowances; a real project derives them from work records. ## 7. Manage shared errors The same software update or a unit error in a shared database can affect many assets simultaneously, even when individual sensors are accurate. Separate versions and rollout waves, verify an independent observation channel and test restoration of local control. Keep each node’s version, change date, owner and known limits in the network log. During an incident, identify the affected cohort and preserve evidence before correction. Explain how verification changed so the same error class is detected before another rollout. ## 8. Turn field experience into a school A successful transfer becomes a new case with raw data, adaptation and outcome. Describe failed transfers with equal care. For every applicability claim, identify tested asset types and remaining gaps. The next practitioner receives the book, data and new tasks. Their agent tests transfer to an asset absent from the examples. Teaching competence rests on repeatable results and an ability to explain limits. Update Academy materials through versioned releases, preserving the history of lessons from implemented projects. ## Calculation laboratory Three asset groups differ in size and performance. Change the portfolio mix: the simple average of percentages and total energy performance will change differently. https://x5power.com/Academy/Hypernetwork/asset-to-network/?lang=en#lab Group sizes and percentages apply per asset. Multiplying sets shows linear workload under unchanged assumptions only. Payback excludes discounting and other expenses. Real expansion requires common-cause failure, seasonality, team-capacity and site-constraint checks. ## Exercises 1. Weighted result Reproduce N-01 baseline energy, savings and portfolio percentage. **Method and reference result** 2,200 MWh; 260 MWh; approximately 11.82%. ## Exercises 2. Mean-percentage error Calculate the mean asset percentage and explain its difference from the portfolio rate. **Method and reference result** 17%. Assets have different baseline consumption, so energy savings use energy weights. ## Exercises 3. Add another large asset Add a second C asset with the same characteristics. **Method and reference result** Baseline 3,200 MWh; savings 310 MWh; rate about 9.69%. Absolute savings grow while the percentage falls. ## Exercises 4. Support cost At €0.18/kWh and €600 support per asset, calculate the annual balance. **Method and reference result** €46,800 gross value and €40,800 after support, before other costs. ## Exercises 5. Zero failures After 20 independent identical trials there are no failures. Calculate the stated upper probability bound. **Method and reference result** 1 − 0.05^(1/20) ≈ 13.91%. Independence and identical distribution are assumptions of this conclusion. ## Exercises 6. Scale and workload Scale the original portfolio by 100. How many commissioning and annual support hours are needed at the stated allowances? **Method and reference result** 1,000 assets; 16,000 commissioning hours; 24,000 annual support hours. Staffing requires available productive hours and seasonal workload distribution. ## Field project Take the result of one workshop and prepare transfer to at least three different assets. Describe the typology, compatibility criteria, new data, calibration, delivery cost, support and rollout stop conditions. Submit a weighted portfolio forecast and workload estimate. Practical defence includes replication on another asset and analysis of a material difference from the pilot. For a larger network, add deployment waves, independent verification and a shared-error scenario. The portfolio model remains a forecast until outcomes are observed at the claimed scale. ## How the work is assessed Academy teaching rubric: 100 points maximum. The calculation defence requires at least 80 points and all mandatory checks. Practical assessment additionally requires an implemented project and measurements. This page prepares the work; it does not issue certificates automatically. - Problem and boundaries: 20. Asset, useful service, units and constraints are defined. - Calculation and data: 25. The main result is reproducible and inputs are traceable. - Result verification: 20. Measurement, comparison and uncertainty match the claim. - Adverse conditions: 15. Failure, changed conditions and recovery are examined. - Operational handover: 10. Owners, acceptance programme and actionable instructions are present. - Transfer to a new asset: 10. New conditions and required adaptations are demonstrated. ### Mandatory checks Units and boundaries are consistent. Every material input has a source. Assumptions and measurements are identified separately. An adverse scenario is tested. The submitted result is reproducible. Practical claims are backed by implemented work. ## Project workbook - Asset and useful service - Boundary, period and units - Inputs and their provenance - Design and alternatives considered - Calculation and testable result - Uncertainty and missing information - Adverse scenario and recovery - Acceptance, owners and follow-up ## Agent assignment Study the book and prerequisites. Reproduce the reference calculation. Solve the exercises before reading the methods. Prepare a project for a new asset. Return inputs with units and provenance, calculation, applicability limits, an adverse scenario, a measurement plan and questions for the human. Give the package to another agent for independent checking. ## Books and primary sources - [EVO · International Performance Measurement and Verification Protocol](https://evo-world.org/en/products-services-mainmenu-en/protocols/ipmvp). Measurement boundaries, baseline and a savings verification plan. Record the chosen protocol edition when applying it. - [NASA · Systems Engineering Handbook](https://www.nasa.gov/reference/systems-engineering-handbook/). Requirements, life cycle, system verification and decision traceability. - [NIST · AI Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework). Evaluation and oversight of AI systems. Pin the edition: AI RMF 1.0 was published in 2023; the site reports an ongoing revision. ## After implementation After one month, check metering and operating conditions. After a season changes, review the forecast and limits. Verify the final claim over its complete stated period. Add new observations to the project history and use them to develop the next exercise. XPower Academy · Alex Ananin and the Echelon group