NSF Mid-Scale Infrastructure Strategy Taking Shape

Share This

Share/Save
Publication date: 
24 October 2018
Number: 
119

The National Science Foundation announced on Oct. 15 that it plans to issue new agency-wide funding opportunities for research infrastructure projects in the range of $6 million to $70 million. The next day, the National Science Board released a report that documents how such mid-scale projects are “underrepresented” in NSF’s portfolio and endorses an agency-wide approach to funding them.

On Oct. 15, NSF announced it plans to issue two agency-wide funding solicitations this fall for mid-scale research infrastructure (RI) proposals. Pending the availability of funds, the agency expects one will be for projects in the range of approximately $6 million to $20 million and the other will be for projects between approximately $20 million to $70 million.

Shortly following NSF’s announcement, its governing body, the National Science Board, released a congressionally mandated study that documents individual NSF research divisions’ struggle to support mid-scale RI. In the report, NSB endorses the creation of long-term agency-wide programs for mid-scale RI and outlines mechanisms for weighing such projects against other major investments.

isr_radar_h.jpg

Advanced Modular Incoherent Scatter Radar (AMISR)

Advanced Modular Incoherent Scatter Radar (AMISR) facilities are designed to make observations of the Earth's ionosphere and space weather events. The Resolute Bay Incoherent Scatter Radar in Canada, pictured above, is a mid-scale project supported by NSF's Division of Atmospheric and Geospace Sciences.

(Image credit - Craig Heinselman / SRI / NSF)

RFI reveals demand for mid-scale projects

NSF defines mid-scale RI as projects whose total cost falls in the gap between its Major Research Instrumentation (MRI) program, which supports the acquisition or development of single instruments for shared use, and its Major Research Equipment and Facilities Construction (MREFC) account, which is primarily used to construct flagship research facilities. Depending on its total appropriation, the MRI program has an upper threshold of $6 million or $9 million (including grantee cost sharing). The MREFC account has a lower threshold of $70 million.

Mid-scale RI can encompass construction projects and upgrades to NSF’s existing large-scale research facilities, as well as the acquisition of instrumentation and equipment, including cyberinfrastructure. Unlike projects funded through the MRI and MREFC programs, these projects are funded through individual research divisions. However, NSF has acknowledged it has proven difficult to fund such projects at the division level because of the relatively large investments they require over a short period of time. 

To assess the research community’s interest in projects within this cost range, NSF asked the research community to submit ideas for mid-scale RI projects across all disciplines through a request for information (RFI) last year. Based on the 192 responses received, NSF determined there is at least $3 billion worth of demand for “high impact” projects in the $20 million to $70 million range.

The Mathematical and Physical Sciences and Geophysics Directorates together accounted for 80 percent of the high impact projects. Among the submissions for those directorates were ideas for particle beam line upgrades, telescopes, and neutrino observatories as well as instrumentation for studying the physical properties of the ocean and atmosphere.

To address this demand, NSF has already requested $60 million in fiscal year 2019 to launch an agency-wide program for mid-scale RI as one of its 10 “Big Ideas.”

Mid-scale infrastructure ‘underrepresented’ in NSF portfolio

The new NSB report adds further support to NSF’s characterization of the difficulties with mid-scale RI. It finds that, although NSF has spent substantial funds on RI generally, mid-scale RI has been “underrepresented” in the agency’s portfolio relative to other investments, such as research centers and large facilities operations and maintenance (O&M).

To characterize NSF’s spending on RI, the board assessed the sizes of awards issued over the past ten years in three funding ranges. It found that mid-scale RI and cyberinfrastructure accounted for a third of awards in the $10 million to $20 million range, 11 percent in the $20 million to $40 million range, and 9 percent in the $40 million to $70 million range.

The board points to the absence of an agency-wide program for mid-scale RI as a potential explanation for the low number of awards, especially in the $20 million to $70 million range. It suggests NSF could partially address this gap by issuing solicitations that provide “clear, consistent signaling to the research community that NSF welcomes such proposals.”

Another contributing factor the board identifies is that mid-scale projects can require significant upfront commitments, creating “an uneven funding profile that meshes poorly with NSF’s investment patterns.” This can result in divisions favoring projects that have more flexible funding requirements, such as research centers. 

Division-level budget tradeoffs can also be a factor, the board notes, pointing to the increasing pressure of rising O&M costs and community inclination to prioritize individual research grants as reasons divisions may be wary of mid-scale investments.

Mid-scale projects lack ‘well-defined home’

While some NSF divisions have dedicated programs for mid-scale RI, the board finds their awards typically only reach the lower part of the mid-scale funding range.

It further observes that a lack of consistent funding has led to a “pervasive perception that NSF does not have a well-defined ‘home’ for mid-scale research infrastructure projects.” Based on interviews with NSF officials, the board finds the limitations in current funding are not due to a lack of interest, but rather are “an artifact of budget constraints and the natural limits of what division budgets can sustain."

The board recounts how one division director said it has become difficult to gauge community interest because, “after years of hearing that NSF does not have the resources to support mid-scale research infrastructure, researchers have stopped enquiring.”

Given NSF’s demonstration of community interest through its RFI, the board endorses the creation of an agency-wide account that initially dedicates $60 million to $100 million per year to mid-scale projects in the $20 million to $70 million range. It also expresses support for NSF’s current exploration of creating a Mid-scale Research Instrumentation Program that would fund projects between the MRI upper threshold and $20 million, emphasizing it “strongly believes” the agency needs to address the effects of inflation on the purchasing power of MRI awards.

The board writes that it believes the agency’s initial request of $60 million represents “a realistic path for NSF to begin mid-scale research infrastructure investments in a restricted funding environment.”

Major facilities account is a potential funding source

NSB was also directed by Congress to consider possible approaches for funding mid-scale RI projects through the MREFC account.

The board finds that using the account in this manner would “raise the strategic priority” of mid-scale RI and increase NSF’s ability to prioritize projects across the agency. It also notes that mid-scale RI projects can lead to future major research facilities and thereby support the MREFC account’s general purpose of supporting large capital investments.

However, the board emphasizes that NSF should be sensitive to the overall composition of its investment portfolio and take care not to undermine the MREFC account’s role in funding major multi-user facilities.  

It also observes that using the account for this purpose would require changes to NSF’s authorizing statutes and the development of oversight procedures that scale to the risk profiles of each project.

Furthermore, given the challenges of weighing a single project in the mid-scale cost range against a potential large facility investment, the board suggests that several mid-scale RI projects could be grouped into a “bundle.” The NSF director would determine the size of the bundle and its priority relative to other MREFC projects, and the board would vote to approve all the proposed mid-scale projects with at once.

While the MREFC account is not the only funding option for mid-scale RI, the board states that “using it for this purpose aligns with NSB’s opinion that a centrally-managed account is key to ensuring agency-level strategic planning and management of the infrastructure portfolio.”

About the author

awolfe [at] aip.org
+1 301-209-3182